Ryzen keeps winning: AMD reaches 33.6% desktop share as Intel slips again

[HardReset]

"Decent performance yes but high performance?" - why the question? you have the benchmarks, and these confirm what I said.

"Again, you are comparing SOC vs CPU" - and? why would I make a distinction when I'm looking at just the CPU performance? FYI just the base m5 is limited to 32GB, the Pro and Max versions can support more. The M5 Max is expected to support 256GB.

"Hint, not using AVX-512 with Zen5 makes that comparison invalid." - that would make Intel vs AMD a moot point too which makes zero sense.

Because you are looking at CPU+memory performance, not CPU performance. Same with video cards. If you want to compare GPUs, you should have somewhat equal:

- Die size
- Power consumption
- Board size
- Memory amount
- Memory interface
- Memory type

etc etc.

Otherwise you are comparing video cards, not GPUs. 256GB, well, Threadrippers support 2TB. Totally different ballpark.

How that makes it invalid? Take Zen5 vs Raptor Lake. Both have same ISA support (Raptor Lake DOES support AVX512, Intel decision is not to use it), both have platform where you can swap CPU, both support DDR5 memory, both are high performance desktop parts.... That is somewhat fair comparison.

I gave you R24 as just a single example, do you want more? I use Handbreak a lot and the M5 beats the 370 HX by about 20% while using almost half the power draw (and power draw is up a lot this generation compared to the M4). This is around what the 9800x3D can do.

Did you use M5 GPU? If yes, you are not comparing CPUs again.

"Again, ARM SOCs do not allow memory expansion like x86-64 CPUs do" - again, we are not comparing platforms. it makes zero difference when comparing raw CPU performance. and forgoing upgradability is a choice made by the OEMs, not a limitation of ARM.

I understand being against Apple or another company (for example I don't buy any laptops that I can't upgrade with more RAM and storage), but why are you so up in arms (pun intended) against ARM as an architecture?

We are comparing platforms. Because ARM "platform" does not support DDR5 or swappable CPU, comparisons are therefore invalid. You can compare server CPU against mobile CPU but that hardly makes any valid comparison. Are there any high performance ARM CPU that support standard DDR5 DIMMs? If not, then it's not OEM choice. And even it is, comparison is still invalid.

What there is to be excited about current ARM architecture solutions? Let's see. Having no legacy instruction support makes it easier to make fast CPU. Having integrated memory is faster than using expandable memory. Small memory is easier to make fast than big one. Having non-swappable CPU makes it easier to lower power consumption. Yeah? All of those are Very Obvious things. ARM has basically done things that are very basic stuff. Nothing special.

As for ARM being faster than x86-64, ARM is not. This is very simple. If ARM CPUs really are faster, then any ARM manufacturer could release platform that is comparable to current x86-64 desktop platforms. Then we would have direct comparison. But because ARM manufacturers know they have no chance, they deliberately avoid fair fight. Just basics of competing. If you have no chance, avoid fair fight.

As for Apple, CPUs being "faster than x86-64". Well, how many desktop class CPUs Apple has released with DDR4 or DDR5 memory support? Zero? That is because if they do, it immediately gives much more fair comparison against others. Because Apple is 100% hype company, Apple avoids direct comparison at all costs. This way Apple can "claim" they have high performance but as usual, that performance has not so much to do with actual CPU performance.

 

[guillejp]

So again, what is the market share of both on desktops, not counting laptops or Steam Deck and portables?

 

[Squid Surprise]

So again, what is the market share of both on desktops, not counting laptops or Steam Deck and portables?

It’s about 68% Intel, 32% AMD… so virtually the same…

 

[Puiu]

Because you are looking at CPU+memory performance, not CPU performance. Same with video cards. If you want to compare GPUs, you should have somewhat equal:

- Die size
- Power consumption
- Board size
- Memory amount
- Memory interface
- Memory type

etc etc.

Otherwise you are comparing video cards, not GPUs. 256GB, well, Threadrippers support 2TB. Totally different ballpark.

How that makes it invalid? Take Zen5 vs Raptor Lake. Both have same ISA support (Raptor Lake DOES support AVX512, Intel decision is not to use it), both have platform where you can swap CPU, both support DDR5 memory, both are high performance desktop parts.... That is somewhat fair comparison.

Did you use M5 GPU? If yes, you are not comparing CPUs again.

We are comparing platforms. Because ARM "platform" does not support DDR5 or swappable CPU, comparisons are therefore invalid. You can compare server CPU against mobile CPU but that hardly makes any valid comparison. Are there any high performance ARM CPU that support standard DDR5 DIMMs? If not, then it's not OEM choice. And even it is, comparison is still invalid.

What there is to be excited about current ARM architecture solutions? Let's see. Having no legacy instruction support makes it easier to make fast CPU. Having integrated memory is faster than using expandable memory. Small memory is easier to make fast than big one. Having non-swappable CPU makes it easier to lower power consumption. Yeah? All of those are Very Obvious things. ARM has basically done things that are very basic stuff. Nothing special.

As for ARM being faster than x86-64, ARM is not. This is very simple. If ARM CPUs really are faster, then any ARM manufacturer could release platform that is comparable to current x86-64 desktop platforms. Then we would have direct comparison. But because ARM manufacturers know they have no chance, they deliberately avoid fair fight. Just basics of competing. If you have no chance, avoid fair fight.

As for Apple, CPUs being "faster than x86-64". Well, how many desktop class CPUs Apple has released with DDR4 or DDR5 memory support? Zero? That is because if they do, it immediately gives much more fair comparison against others. Because Apple is 100% hype company, Apple avoids direct comparison at all costs. This way Apple can "claim" they have high performance but as usual, that performance has not so much to do with actual CPU performance.

You are making zero sense dude.

"Did you use M5 GPU?" - This is a question that you put out when you lose all logical arguments. Are we still talking about the CPU? Or are just trying to take a piss at me with some wild gaslighting? We are talking about CPU performance and you come at me with "but our GPUs are better!" and I'm supposed to take you seriously?

"If ARM CPUs really are faster, then any ARM manufacturer could release platform that is comparable to current x86-64 desktop platforms." - Is Apple not part of that "manufacturer" list of yours? And we did have other manufacturers release systems with ARM, the problem wasn't the performance, it was the fledgeling software environment which had compatibility issues running some non-native applications and higher than expected prices.

"Because ARM "platform" does not support DDR5 or swappable CPU, comparisons are therefore invalid." - What is invalid is the way you try to move the goal post. The platform doesn't matter. You have x86 platforms with the same RAM used by ARM which aren't replaceable. Didn't intel even do RAM on SOC recently with Lunar Lake?

"Well, how many desktop class CPUs Apple has released with DDR4 or DDR5 memory support?" - nobody cares and it's not what we were talking about. You can blame Intel and AMD for refusing to use all of the available tools they have available (AMD for example refused to upgrade the memory controller on desktop PC this gen). if PC is the land of freedom, then we should be allowed to choose the type of RAM we use too (some would 100% buy soldered RAM if it gave a tangible boost in performance).

 

[HardReset]

You are making zero sense dude.

"Did you use M5 GPU?" - This is a question that you put out when you lose all logical arguments. Are we still talking about the CPU? Or are just trying to take a piss at me with some wild gaslighting? We are talking about CPU performance and you come at me with "but our GPUs are better!" and I'm supposed to take you seriously?

That was a question. See, video encoding/decoding/editing etc can be done with GPU too. And Handbrake does support GPU. And quick web search tells Handbrake is very slow on Apple hardware. Feel free to search yourself and see.

I asked a simple question you refuse to answer ""

"If ARM CPUs really are faster, then any ARM manufacturer could release platform that is comparable to current x86-64 desktop platforms." - Is Apple not part of that "manufacturer" list of yours? And we did have other manufacturers release systems with ARM, the problem wasn't the performance, it was the fledgeling software environment which had compatibility issues running some non-native applications and higher than expected prices.

"Because ARM "platform" does not support DDR5 or swappable CPU, comparisons are therefore invalid." - What is invalid is the way you try to move the goal post. The platform doesn't matter. You have x86 platforms with the same RAM used by ARM which aren't replaceable. Didn't intel even do RAM on SOC recently with Lunar Lake?

Ah, performance is not problem. Again, we are talking about CPU performance. Then everything else than CPU alone should be left out. And that includes memory performance, SSD performance etc etc.

You obviously didn't read my GPU example where it is very clear that comparing GPU performance mean comparing GPUs only, not everything else outside GPU. To compare CPU performance everything outside CPU cores (CPU caches are OK if they are with CPU) should be somewhat equal. Then CPU comparison is fair. Otherwise it's more platform comparison or something else. Feel free to post benchmarks that rely very little on memory performance. That makes comparison at least somewhat fair.

AMD and Intel have solutions with integrated memory but most CPUs are still with expandable memory and have been designed for that. Because ARM solutions don't offer same options, comparison is again more platform and less CPU.

"Well, how many desktop class CPUs Apple has released with DDR4 or DDR5 memory support?" - nobody cares and it's not what we were talking about. You can blame Intel and AMD for refusing to use all of the available tools they have available (AMD for example refused to upgrade the memory controller on desktop PC this gen). if PC is the land of freedom, then we should be allowed to choose the type of RAM we use too (some would 100% buy soldered RAM if it gave a tangible boost in performance).

I care because that tells Apple does not want anyone to compare their CPUs against AMD and Intel. PC is land of freedom, yes. Freedom is your ability to choose howe much memory you want and upgrade/switch it later. Apple gives you zero freedom. You are stuck with memory amount you buy. If it's 8 GB, then good luck. Basically you tell that AMD and Intel should take freedom to choose what memory and how much and that is freedom? Well, they are going on that route too but again, that is less freedom then. And that is why they have been hesitant to do it so long.

 

[Puiu]

That was a question. See, video encoding/decoding/editing etc can be done with GPU too. And Handbrake does support GPU. And quick web search tells Handbrake is very slow on Apple hardware. Feel free to search yourself and see.

I asked a simple question you refuse to answer ""

Ah, performance is not problem. Again, we are talking about CPU performance. Then everything else than CPU alone should be left out. And that includes memory performance, SSD performance etc etc.

You obviously didn't read my GPU example where it is very clear that comparing GPU performance mean comparing GPUs only, not everything else outside GPU. To compare CPU performance everything outside CPU cores (CPU caches are OK if they are with CPU) should be somewhat equal. Then CPU comparison is fair. Otherwise it's more platform comparison or something else. Feel free to post benchmarks that rely very little on memory performance. That makes comparison at least somewhat fair.

AMD and Intel have solutions with integrated memory but most CPUs are still with expandable memory and have been designed for that. Because ARM solutions don't offer same options, comparison is again more platform and less CPU.

I care because that tells Apple does not want anyone to compare their CPUs against AMD and Intel. PC is land of freedom, yes. Freedom is your ability to choose howe much memory you want and upgrade/switch it later. Apple gives you zero freedom. You are stuck with memory amount you buy. If it's 8 GB, then good luck. Basically you tell that AMD and Intel should take freedom to choose what memory and how much and that is freedom? Well, they are going on that route too but again, that is less freedom then. And that is why they have been hesitant to do it so long.

"And quick web search tells Handbrake is very slow on Apple hardware." - When I told the fact that an M5 is similar in handbrake to an 9800x3d, you simply ignored it. (the 9800x3d is a CPU I have at home and I'm also using an M4 Max at work) As for "video" everybody knows that Apple is the way to go unless you have some very specific requirements. Ask any content creator.

"Then everything else than CPU alone should be left out. And that includes memory performance, SSD performance etc etc." - yes, it's completely irrelevant. you tried to tell me that the ARM CPU is not fast enough and now that I've proven that it is, you are clinging to "platform" as if that makes any difference in performance argument. if you can't buy an x86 CPU with the RAM used by Apple then that's a platform issue which you should acknowledge, don't use it as an excuse.

We went from arguing about raw CPU speed and power draw to you telling me that, because a PC can get a better GPU ,the x86 CPU is better and that it's unfair that Apple is using soldered memory... what exactly are you trying to accomplish dude?

And just because you like to talk about GPUs, the M5 iGPU is about as fast as the Intel Arc 140V iGPU (aka Xe² Battlemage) in games that are available on the mac. For other things the M5 is much faster.
And to put things into perspective, the M4 Max is close to the RTX 4080 (desktop) in blender render tests (Metal vs Optix) so this should put into perspective what the future M5 Max will be like.
I've seen people test the M5 in blender and it was just behind a 3070ti laptop.

Hell, the M5 was proven to be faster in single core performance than any x86 CPU while running windows 11 in a virtual environment... that should tell you everything you need to know about the current ARM architecture.

You can bash Apple for it's crappy restrictions all you want (I know I do every single time I get the chance), but do not turn a blind eye to what the performance their hardware can achieve. It helps nobody if you do that.

 

[HardReset]

"And quick web search tells Handbrake is very slow on Apple hardware." - When I told the fact that an M5 is similar in handbrake to an 9800x3d, you simply ignored it. (the 9800x3d is a CPU I have at home and I'm also using an M4 Max at work) As for "video" everybody knows that Apple is the way to go unless you have some very specific requirements. Ask any content creator.

"Then everything else than CPU alone should be left out. And that includes memory performance, SSD performance etc etc." - yes, it's completely irrelevant. you tried to tell me that the ARM CPU is not fast enough and now that I've proven that it is, you are clinging to "platform" as if that makes any difference in performance argument. if you can't buy an x86 CPU with the RAM used by Apple then that's a platform issue which you should acknowledge, don't use it as an excuse.

I didn't ignore it. I asked if you used also GPU, since web searches tell Apple is slow on Handbrake.

Problem here is: yoyu take small niche scenario and then try to scale it to cover everything. That niche scenario happens to be one where raw CPU performance is not that important. It is strange that ARM CPUs avoid competing In scenario where raw CPU power matters most if ARM really has that much power.

We went from arguing about raw CPU speed and power draw to you telling me that, because a PC can get a better GPU ,the x86 CPU is better and that it's unfair that Apple is using soldered memory... what exactly are you trying to accomplish dude?

And just because you like to talk about GPUs, the M5 iGPU is about as fast as the Intel Arc 140V iGPU (aka Xe² Battlemage) in games that are available on the mac. For other things the M5 is much faster.
And to put things into perspective, the M4 Max is close to the RTX 4080 (desktop) in blender render tests (Metal vs Optix) so this should put into perspective what the future M5 Max will be like.
I've seen people test the M5 in blender and it was just behind a 3070ti laptop.

Apphe has not swappable CPU and expandable memory. Having both things mean harder to have high performance platform. That is, comparison is More platform than CPU alone. Like it or not, that is how it is.

CPU integrated GPU is easily faster and More energy efficient than external one. Everyone knows that. Nothing special there.

Hell, the M5 was proven to be faster in single core performance than any x86 CPU while running windows 11 in a virtual environment... that should tell you everything you need to know about the current ARM architecture.@@

Again, memory subsystem plays role there. Niche scenario etc (like told above). How about giving Apple benchmarks with CPU that has DDR5 support and swappable CPU at least?

You can bash Apple for it's crappy restrictions all you want (I know I do every single time I get the chance), but do not turn a blind eye to what the performance their hardware can achieve. It helps nobody if you do that.

Apple can achieve performance on niche scenario. That is veeery far from achieving performance outside that niche scenario. There has been countless x86-64 killer CPUs but so far no one has really challenged AMD AND Intel.

 

[Puiu]

I didn't ignore it. I asked if you used also GPU, since web searches tell Apple is slow on Handbrake.

Problem here is: yoyu take small niche scenario and then try to scale it to cover everything. That niche scenario happens to be one where raw CPU performance is not that important. It is strange that ARM CPUs avoid competing In scenario where raw CPU power matters most if ARM really has that much power.

Apphe has not swappable CPU and expandable memory. Having both things mean harder to have high performance platform. That is, comparison is More platform than CPU alone. Like it or not, that is how it is.

CPU integrated GPU is easily faster and More energy efficient than external one. Everyone knows that. Nothing special there.

Again, memory subsystem plays role there. Niche scenario etc (like told above). How about giving Apple benchmarks with CPU that has DDR5 support and swappable CPU at least?

Apple can achieve performance on niche scenario. That is veeery far from achieving performance outside that niche scenario. There has been countless x86-64 killer CPUs but so far no one has really challenged AMD AND Intel.

"Problem here is: yoyu take small niche scenario and then try to scale it to cover everything." - I gave you plenty of examples. you are the one who wants very specific results. give me examples of some CPU benchmarks that you want me to research and that you will accept.

"Apphe has not swappable CPU and expandable memory. " - and? it's irrelevant. we are talking about CPU architectures and how fast ARM is today, not Apple's business decisions.

"CPU integrated GPU is easily faster and More energy efficient than external one." - I never once compared GPU power usage, just CPU.

"Apple can achieve performance on niche scenario." - ok, show me general results.

Tell me that these results are bad (albeit, expensive for a laptop ) and add 15-30% for the not yet released M5 Max:

 

[HardReset]

"Problem here is: yoyu take small niche scenario and then try to scale it to cover everything." - I gave you plenty of examples. you are the one who wants very specific results. give me examples of some CPU benchmarks that you want me to research and that you will accept.

"Apphe has not swappable CPU and expandable memory. " - and? it's irrelevant. we are talking about CPU architectures and how fast ARM is today, not Apple's business decisions.

"CPU integrated GPU is easily faster and More energy efficient than external one." - I never once compared GPU power usage, just CPU.

You still have niche scenario. Let me illustrate:

Take two identical systems except differences told below. CPUs are totally identical, say 9800X3D for example. Because CPUs are totally identical, CPU performance on benchmarks should also be identical, right?

- System A has fastest DDR5 memory available, system B has slowest DDR5 memory available.
- Same as above but benchmark use 64 GB RAM, system A has 16 GB and system B has 96 GB.
- Same as above but both systems have 16 GB RAM, system A has PCIe 5.0 NVMe SSD and system B has SATA HDD 5400 RPM.

Identical CPUs, identical CPU performance? Like I said long time ago, to compare CPU performance, everything else outside CPU must be somewhat equal. Otherwise you are easily comparing something else than CPU performance.

You talk CPU architectures but forget to add memory subsystem into equation?

You said Apple GPU vs many things. Fyi, power consumption matters on GPUs a lot.

"Apple can achieve performance on niche scenario." - ok, show me general results.

Easy, show me benchmarks using Apple CPU that use standard DDDR5 DIMM and use socket where you can switch CPU. Then we are talking.

Tell me that these results are bad (albeit, expensive for a laptop ) and add 15-30% for the not yet released M5 Max:

I gladly put that machine against my rig. Just run benchmark that use 64 GB RAM and I pretty much guarantee my 96 GB rig obliterates it.

He’ll never stop… he’s an AMD shill… pretty sure even he knows he’s arguing a lost cause but - kudos for the effort…

As usual, you are trolling because that's all you can do.

 

[Squid Surprise]

You still have niche scenario. Let me illustrate:

Take two identical systems except differences told below. CPUs are totally identical, say 9800X3D for example. Because CPUs are totally identical, CPU performance on benchmarks should also be identical, right?

- System A has fastest DDR5 memory available, system B has slowest DDR5 memory available.
- Same as above but benchmark use 64 GB RAM, system A has 16 GB and system B has 96 GB.
- Same as above but both systems have 16 GB RAM, system A has PCIe 5.0 NVMe SSD and system B has SATA HDD 5400 RPM.

Identical CPUs, identical CPU performance? Like I said long time ago, to compare CPU performance, everything else outside CPU must be somewhat equal. Otherwise you are easily comparing something else than CPU performance.

You talk CPU architectures but forget to add memory subsystem into equation?

You said Apple GPU vs many things. Fyi, power consumption matters on GPUs a lot.

Easy, show me benchmarks using Apple CPU that use standard DDDR5 DIMM and use socket where you can switch CPU. Then we are talking.

I gladly put that machine against my rig. Just run benchmark that use 64 GB RAM and I pretty much guarantee my 96 GB rig obliterates it.

As usual, you are trolling because that's all you can do.

lol… you only extol AMD and ignore actual facts… I think we know who the real troll is…

 

[Gezzer]

Which covers 99.9% of all corporate or non-gamer offerings.

That's the big thing. It's the same reason that M$ makes windows more for enterprise customers than everyone else. Gamers and hobbyists are a drop in the bucket in comparison. It's also why Nvidia pivoted to AI over gamer cards.

 

[Puiu]

You still have niche scenario. Let me illustrate:

Take two identical systems except differences told below. CPUs are totally identical, say 9800X3D for example. Because CPUs are totally identical, CPU performance on benchmarks should also be identical, right?

- System A has fastest DDR5 memory available, system B has slowest DDR5 memory available.
- Same as above but benchmark use 64 GB RAM, system A has 16 GB and system B has 96 GB.
- Same as above but both systems have 16 GB RAM, system A has PCIe 5.0 NVMe SSD and system B has SATA HDD 5400 RPM.

Identical CPUs, identical CPU performance? Like I said long time ago, to compare CPU performance, everything else outside CPU must be somewhat equal. Otherwise you are easily comparing something else than CPU performance.

You talk CPU architectures but forget to add memory subsystem into equation?

You said Apple GPU vs many things. Fyi, power consumption matters on GPUs a lot.

Easy, show me benchmarks using Apple CPU that use standard DDDR5 DIMM and use socket where you can switch CPU. Then we are talking.

I gladly put that machine against my rig. Just run benchmark that use 64 GB RAM and I pretty much guarantee my 96 GB rig obliterates it.

As usual, you are trolling because that's all you can do.

Your RAM speed obsession is something else. You can use 8000MHz RAM on AMD. Does it make it faster? Not really.

And the results I've given you are not RAM capacity starved so I don't get why you are talking about the amount of RAM.

"Same as above but both systems have 16 GB RAM, system A has PCIe 5.0 NVMe SSD and system B has SATA HDD 5400 RPM." - ok, I admit I've used an HDD from 1996 for the x86 benchmarks.

"You said Apple GPU vs many things. Fyi, power consumption matters on GPUs a lot." - not in the context of our discussion. you only brought up GPUs as an excuse to ignore CPU performance.

"show me benchmarks using Apple CPU that use standard DDDR5 DIMM" - we have x86 results that use the type of RAM Apple uses, you argument makes zero sense. intel went even further and put the RAM on the SOC.

"As usual, you are trolling because that's all you can do." - says the guy who refuses to look at real world results and clings to some fictional x86 superiority story full of "excuses".

 

[HardReset]

lol… you only extol AMD and ignore actual facts… I think we know who the real troll is…

I'm one telling facts here.

Your RAM speed obsession is something else. You can use 8000MHz RAM on AMD. Does it make it faster? Not really.

And the results I've given you are not RAM capacity starved so I don't get why you are talking about the amount of RAM.

"Same as above but both systems have 16 GB RAM, system A has PCIe 5.0 NVMe SSD and system B has SATA HDD 5400 RPM." - ok, I admit I've used an HDD from 1996 for the x86 benchmarks.

Faster memory makes system work faster. Imagine AMD vs Intel test where one uses slower memory than supported. Outcry is guaranteed.

Because you deliberately cherry pick scenarios where Amount of RAM is not important but Speed of RAM is. That is, because currently Apple cannot match amount of RAM and AMD cannot match speed of RAM. So you cherry pick scenario that favour Apple and then say Apple has faster CPU

Obviously you miss point that storage speed could also matter there.

"You said Apple GPU vs many things. Fyi, power consumption matters on GPUs a lot." - not in the context of our discussion. you only brought up GPUs as an excuse to ignore CPU performance.

"show me benchmarks using Apple CPU that use standard DDDR5 DIMM" - we have x86 results that use the type of RAM Apple uses, you argument makes zero sense. intel went even further and put the RAM on the SOC.

"As usual, you are trolling because that's all you can do." - says the guy who refuses to look at real world results and clings to some fictional x86 superiority story full of "excuses".

Your "context" is CPU speed architecture wise. Still you ignore everything Outside CPU.

There are Some x86 SOCs with integrated memory, still all benchmarks you show are about those CPUs without it. Like your video post. It also matters how that memory bandwidth is distributed between CPU and GPU. No to mention those integrated x86 based solutions currently have low core counts. Your arguments here make no sense.

YOU are talking about CPU architecture performance. Then you deliberately avoid situations where we can actually compare CPU architecture performance. Instead you take scenario where amount of memory matters very little but speed of memory matters a lot. That just happens to be scenario where Apple SOCs look as good as possible. YOU are one making excuses here.

How about running benchmark that requires 64 GB RAM on your 16 GB Apple chip? It's not AMD fault that Apple pust so small amount of RAM.

 

[Puiu]

I'm one telling facts here.

Faster memory makes system work faster. Imagine AMD vs Intel test where one uses slower memory than supported. Outcry is guaranteed.

Because you deliberately cherry pick scenarios where Amount of RAM is not important but Speed of RAM is. That is, because currently Apple cannot match amount of RAM and AMD cannot match speed of RAM. So you cherry pick scenario that favour Apple and then say Apple has faster CPU

Obviously you miss point that storage speed could also matter there.

Your "context" is CPU speed architecture wise. Still you ignore everything Outside CPU.

There are Some x86 SOCs with integrated memory, still all benchmarks you show are about those CPUs without it. Like your video post. It also matters how that memory bandwidth is distributed between CPU and GPU. No to mention those integrated x86 based solutions currently have low core counts. Your arguments here make no sense.

YOU are talking about CPU architecture performance. Then you deliberately avoid situations where we can actually compare CPU architecture performance. Instead you take scenario where amount of memory matters very little but speed of memory matters a lot. That just happens to be scenario where Apple SOCs look as good as possible. YOU are one making excuses here.

How about running benchmark that requires 64 GB RAM on your 16 GB Apple chip? It's not AMD fault that Apple pust so small amount of RAM.

"Imagine AMD vs Intel test where one uses slower memory than supported." - I don't need to imagine anything. We have multiple comparison benchmarks with RAM from 4800MHz to 8000MHz.

here you go:

Is DDR5-8000 Worth It? The Ryzen AM5 Test

Thinking about upgrading to AMD's AM5 platform? We've tested nearly everything from CPUs to motherboards---now it's time to revisit DDR5 memory performance, including G.Skill's new CL26 and...

www.techspot.com

DDR5 Memory Performance Scaling with AMD Zen 5

Our AMD Zen 5 Memory Scaling Review examines six different DDR5 memory speeds, including DDR5-8000, DDR5-6400 1:1, DDR5-6400 CL28 and others. We looked into how these speeds impact performance in gaming at four resolutions and a wide range of application workloads.

www.techpowerup.com

DDR5 6000 vs 8000 MHz RAM - AMD Ryzen 9 9950X - Hardware Busters

Hardware Busters - DDR5 6000 vs 8000 MHz RAM - AMD Ryzen 9 9950X - CPU

hwbusters.com

You can see exactly what the memory bandwidth starved benchmarks/tests are. If you study those 3 links.

"That is, because currently Apple cannot match amount of RAM" - yes it can. you can even get the Ultra systems that support 512GB, but that's more an workstation equivalent. the M4 Max supports 128 and the M5 Max 256.

"still all benchmarks you show are about those CPUs without it." - you have Intel lunar lake and AMD AI MAX reviews online... why are you even mentioning this without searching yourself? (both of those have systems with no upgradeable memory, just like ARM systems)

"Then you deliberately avoid situations where we can actually compare CPU architecture performance." - No I haven't. You just want mythical perfectly equivalent systems that don't exist. And based on the benchmarks shown, even if you had PCs with the same RAM, there will be too little difference in performance to change the conclusion. You are not going to tell me that a 5-12% in 7zip (which is a known memory hungry test) is going to change the overall scores now, are you?

You simply do not like the fact that ARM is actually competitive. You cling to "RAM" speeds and "upgradability" as if that changes the results in any way. You behave as if I've shown you benchmarks of PCs with RAM set to 4800MHz, not their recommended very high end specs.

No amount of copium installed in a PC is going to make x86 gain 30% in ST and cut power draw in half. Not for at least 2 more generations. We have to accept this little known fact.

You sound like you've never actually used a Macbook pro. I have many issues with them, but you can feel how snappy things are compared to windows and linux. The browsers especially feel very fast.

 

[Squid Surprise]

I'm one telling facts here.



Because you deliberately cherry pick scenarios

sorry, had a nice laugh at that… could you maybe reread your posts and come back to us? All you’ve done is cherry pick examples where your believed AMD is better…

 

[HardReset]

"Imagine AMD vs Intel test where one uses slower memory than supported." - I don't need to imagine anything. We have multiple comparison benchmarks with RAM from 4800MHz to 8000MHz.

here you go:

Is DDR5-8000 Worth It? The Ryzen AM5 Test

Thinking about upgrading to AMD's AM5 platform? We've tested nearly everything from CPUs to motherboards---now it's time to revisit DDR5 memory performance, including G.Skill's new CL26 and...

www.techspot.com

DDR5 Memory Performance Scaling with AMD Zen 5

Our AMD Zen 5 Memory Scaling Review examines six different DDR5 memory speeds, including DDR5-8000, DDR5-6400 1:1, DDR5-6400 CL28 and others. We looked into how these speeds impact performance in gaming at four resolutions and a wide range of application workloads.

www.techpowerup.com

DDR5 6000 vs 8000 MHz RAM - AMD Ryzen 9 9950X - Hardware Busters

Hardware Busters - DDR5 6000 vs 8000 MHz RAM - AMD Ryzen 9 9950X - CPU

hwbusters.com

You can see exactly what the memory bandwidth starved benchmarks/tests are. If you study those 3 links.

"That is, because currently Apple cannot match amount of RAM" - yes it can. you can even get the Ultra systems that support 512GB, but that's more an workstation equivalent. the M4 Max supports 128 and the M5 Max 256.

You finally realized that memory speed matters on speed. Excellent.

Those figures are still waaay behind what equivalent AMD systems support. That's one reason AMD sticks with DIMMs.

"still all benchmarks you show are about those CPUs without it." - you have Intel lunar lake and AMD AI MAX reviews online... why are you even mentioning this without searching yourself? (both of those have systems with no upgradeable memory, just like ARM systems)

"Then you deliberately avoid situations where we can actually compare CPU architecture performance." - No I haven't. You just want mythical perfectly equivalent systems that don't exist. And based on the benchmarks shown, even if you had PCs with the same RAM, there will be too little difference in performance to change the conclusion. You are not going to tell me that a 5-12% in 7zip (which is a known memory hungry test) is going to change the overall scores now, are you?

You simply do not like the fact that ARM is actually competitive. You cling to "RAM" speeds and "upgradability" as if that changes the results in any way. You behave as if I've shown you benchmarks of PCs with RAM set to 4800MHz, not their recommended very high end specs.

AMD AI Max has much lower clock speed than Ryzens. In other words, Ryzen AI Max is downgraded Ryzen when it comes to clock speed.

I want "mythical perfectly equivalent systems" because that is standard benchmark procedure when comparing different CPU architectures.

For example Techspot Zen 5 review https://www.techspot.com/review/2877-amd-ryzen-7-9700x/

Other than different motherboard and slightly different memory settings, everything else is equal! That is how CPU architecture benchmarks should be done. It just seems that because Apple way is not possible, then standard procedure suddenly becomes something that don't exist

Apple way is only way ""

Those things have effect on almost everything. Standard procedures for benchmarking exist for good reasons. If you are unable to test fairly, then you should just accept that there is no good way to compare.

At least you now said ARM is (only) competitive. On some areas, I can accept that. Before that it was like ARM CPUs are best ever and x86 have no chance anywhere.

No amount of copium installed in a PC is going to make x86 gain 30% in ST and cut power draw in half. Not for at least 2 more generations. We have to accept this little known fact.

You sound like you've never actually used a Macbook pro. I have many issues with them, but you can feel how snappy things are compared to windows and linux. The browsers especially feel very fast.

Yeah right, just enabling AVX-512 support on software (it takes few seconds to enable flag on compiler) easily gives way over that 30%. You can find many benchmarks on Phoronix like https://www.phoronix.com/review/amd-epyc-turin-avx512

Too bad, for example Cinebench developers are Intel fanboys (they admitted that AVX512 is not supported because some Skylake (2015 and such) era Intels run slower) and now they are probably switched for Apple fanboys. However software optimizations have Nothing to do with CPU architecture speed.

I have never even touched any Apple device. I consider them dangerous and avoid them at all costs.

sorry, had a nice laugh at that… could you maybe reread your posts and come back to us? All you’ve done is cherry pick examples where your believed AMD is better…

I asked for standard benchmark procedures but for Apple fanboys they are of course something unbeliveable and impossile.

 

[Squid Surprise]

I asked for standard benchmark procedures but for Apple fanboys they are of course something unbeliveable and impossile.

I see… so you wish to disregard facts unless they agree with your world view… gotcha!

 

[Puiu]

You finally realized that memory speed matters on speed. Excellent.

Those figures are still waaay behind what equivalent AMD systems support. That's one reason AMD sticks with DIMMs.

AMD AI Max has much lower clock speed than Ryzens. In other words, Ryzen AI Max is downgraded Ryzen when it comes to clock speed.

I want "mythical perfectly equivalent systems" because that is standard benchmark procedure when comparing different CPU architectures.

For example Techspot Zen 5 review https://www.techspot.com/review/2877-amd-ryzen-7-9700x/

Other than different motherboard and slightly different memory settings, everything else is equal! That is how CPU architecture benchmarks should be done. It just seems that because Apple way is not possible, then standard procedure suddenly becomes something that don't exist

Apple way is only way ""

Those things have effect on almost everything. Standard procedures for benchmarking exist for good reasons. If you are unable to test fairly, then you should just accept that there is no good way to compare.

At least you now said ARM is (only) competitive. On some areas, I can accept that. Before that it was like ARM CPUs are best ever and x86 have no chance anywhere.

Yeah right, just enabling AVX-512 support on software (it takes few seconds to enable flag on compiler) easily gives way over that 30%. You can find many benchmarks on Phoronix like https://www.phoronix.com/review/amd-epyc-turin-avx512

Too bad, for example Cinebench developers are Intel fanboys (they admitted that AVX512 is not supported because some Skylake (2015 and such) era Intels run slower) and now they are probably switched for Apple fanboys. However software optimizations have Nothing to do with CPU architecture speed.

I have never even touched any Apple device. I consider them dangerous and avoid them at all costs.

I asked for standard benchmark procedures but for Apple fanboys they are of course something unbeliveable and impossile.

In the end all you did was tell us your personal feelings and ignored all benchmarks, reviews and objective facts.

"Yeah right, just enabling AVX-512 support on software" - show me the applications you used in the past year that give 30% with AVX 512 enabled. Are you running miniBUDE and GROMACS on your home PC?

And better yet, shows us examples where AVX 512 enabled apps perform better than on ARM without AVX 512. The only one I know that can actually give that 30% boost and make a difference is RPCS3 (it got a short while ago ARM64 support so it has beta support ARM MacOS now)

You don't like CB24 because it doesn't have AVX 512? Ok, let me show you Blender (something I've actively use on both Windows and MacOS), which does support AVX 512 when CPU rendering. The M4 Max sits exactly where CB puts it for multithreading, slightly above the 14700k - so all of this just amounted of me wasting time to prove you wrong again. (can you tell me the power draw of those CPUs compared to the M4 Max?)



This will put the M5 Max around the top, alongside AMD and Intel in an AVX 512 enabled benchmark. (probably next to the 285k)

"that is standard benchmark procedure when comparing different CPU architectures" - So when Intel is being benched every time with higher speed memory, you complain? No you don't. You don't even mention AVX either. This is just you having beef with Apple and ARM. I gave you examples where you have equivalent RAM setups and you just ignored them and only kept talking about desktop PC. And when I gave you RAM scaling you took a 5% increase in some benches as a "win" as if that's enough to change anything.

from the review you linked:
AMD 7000/9000 system: G.Skill Trident Z5 RGB 32GB DDR5-6000
AMD 5000 system: G.Skill Ripjaws V Series 32GB DDR4-3600
Intel 12th, 13th & 14th gen system: G.Skill Trident Z5 RGB 32GB DDR5-7200

Incredibly equivalent systems!!!!! It's just slightly different memory settings! Thanks for showing us that we were wrong! /s

"Before that it was like ARM CPUs are best ever" - if you remove the context then everything I say can become a stupid comment. I've always talked about its incredible ST performance and efficiency. And you started telling me that x86 with faster RAM will somehow beat ARM (or get close to it) and you never actually addressed efficiency besides ranting about the GPU that one time. What exactly am I wrong about in that statement?

I also feel the need to remind you of another context since you seem to forgot. It's all in the context of the benchmark results the M5 has shown us and the soon to be released M5 Max. I'm using M4 Max is just to give you reference point of comparison so you can easily infer the performance of the M5 Max

"I asked for standard benchmark" - I gave you standard benchmarks and even gave you non-standard ones for x86 so you can infer the performance in difference memory setups. It's all you need to know to be able to make an informed objective comparison.

 

[HardReset]

I see… so you wish to disregard facts unless they agree with your world view… gotcha!

No, you are one that won't accept clear facts.

In the end all you did was tell us your personal feelings and ignored all benchmarks, reviews and objective facts.

"Yeah right, just enabling AVX-512 support on software" - show me the applications you used in the past year that give 30% with AVX 512 enabled. Are you running miniBUDE and GROMACS on your home PC?

And better yet, shows us examples where AVX 512 enabled apps perform better than on ARM without AVX 512. The only one I know that can actually give that 30% boost and make a difference is RPCS3 (it got a short while ago ARM64 support so it has beta support ARM MacOS now)

You can check Phoronix articles about AVX512 performance boost on Zen5. There are benchmark facts about Zen5 performance gains. You probably ignored those.

Problem is, enabling AVX512 support on software requires recompilation of software (easiest way). Another problem is that developers rarely do it.

You don't like CB24 because it doesn't have AVX 512? Ok, let me show you Blender (something I've actively use on both Windows and MacOS), which does support AVX 512 when CPU rendering. The M4 Max sits exactly where CB puts it for multithreading, slightly above the 14700k - so all of this just amounted of me wasting time to prove you wrong again. (can you tell me the power draw of those CPUs compared to the M4 Max?)

View attachment 90729

This will put the M5 Max around the top, alongside AMD and Intel in an AVX 512 enabled benchmark. (probably next to the 285k)

As you can clearly see, AVX512 improvement is very small (Zen4 vs Zen5). It is either AVX512 support for Blender is bad or Blender is not good software to test AVX512 improvement. Again, you are taking ONE example and try to say that applies to everything. No surprise, it does not. I already showed examples that tell different story.

"that is standard benchmark procedure when comparing different CPU architectures" - So when Intel is being benched every time with higher speed memory, you complain? No you don't. You don't even mention AVX either. This is just you having beef with Apple and ARM. I gave you examples where you have equivalent RAM setups and you just ignored them and only kept talking about desktop PC. And when I gave you RAM scaling you took a 5% increase in some benches as a "win" as if that's enough to change anything.

from the review you linked:
AMD 7000/9000 system: G.Skill Trident Z5 RGB 32GB DDR5-6000
AMD 5000 system: G.Skill Ripjaws V Series 32GB DDR4-3600
Intel 12th, 13th & 14th gen system: G.Skill Trident Z5 RGB 32GB DDR5-7200

Incredibly equivalent systems!!!!! It's just slightly different memory settings! Thanks for showing us that we were wrong! /s

AVX? Both Intel and AMD support AVX, no difference there.

For AMD 5000 series, they do not support DDR5 so they must have DDR4 instead. No, you cannot test CPU that support only DDR4 with DDR5. Also in case Intel OFFICIALLY supports higher memory speeds, higher memory speeds are used. We may argue that DDR5 speeds must be exactly same but again, using highest officially supported memory speed is somewhat OK.

Setup is very close to identical except small difference in memory speeds (DDR5) and when DDR5 cannot be used, DDR4 is used instead.

Not that you take some CPU with integrated memory and integrated GPU because you cannot use external GPU or external memory.

"Before that it was like ARM CPUs are best ever" - if you remove the context then everything I say can become a stupid comment. I've always talked about its incredible ST performance and efficiency. And you started telling me that x86 with faster RAM will somehow beat ARM (or get close to it) and you never actually addressed efficiency besides ranting about the GPU that one time. What exactly am I wrong about in that statement?

I also feel the need to remind you of another context since you seem to forgot. It's all in the context of the benchmark results the M5 has shown us and the soon to be released M5 Max. I'm using M4 Max is just to give you reference point of comparison so you can easily infer the performance of the M5 Max

"I asked for standard benchmark" - I gave you standard benchmarks and even gave you non-standard ones for x86 so you can infer the performance in difference memory setups. It's all you need to know to be able to make an informed objective comparison.

Your examples are two software that make very little use of AVX512, you proved it yourself. That is problem. You also use SOC with integrated memory against CPU with external memory. Your comparison is simply invalid because of those two things. I don't even start to list other things that are wrong because those two are more than enough.

OK, we probably CAN estimate M5 performance based on M4 but this was about ARM vs x86-64, not ARM vs ARM?

I asked for standard benchmark procedures. You still cannot give them and considering everything so far, you cannot give them. How about just admitting that at this point there is no proper way to directly compare Apple products vs current x86-64 CPUs? You are trying to do something that simply cannot be done. And refuse to accept that fact.

 

[Squid Surprise]

No, you are one that won't accept clear facts.

I asked for standard benchmark procedures. You still cannot give them and considering everything so far, you cannot give them. How about just admitting that at this point there is no proper way to directly compare Apple products vs current x86-64 CPUs? You are trying to do something that simply cannot be done. And refuse to accept that fact.

I believe the goalposts moved again…

 

[Puiu]

No, you are one that won't accept clear facts.

You can check Phoronix articles about AVX512 performance boost on Zen5. There are benchmark facts about Zen5 performance gains. You probably ignored those.

Problem is, enabling AVX512 support on software requires recompilation of software (easiest way). Another problem is that developers rarely do it.

As you can clearly see, AVX512 improvement is very small (Zen4 vs Zen5). It is either AVX512 support for Blender is bad or Blender is not good software to test AVX512 improvement. Again, you are taking ONE example and try to say that applies to everything. No surprise, it does not. I already showed examples that tell different story.

AVX? Both Intel and AMD support AVX, no difference there.

For AMD 5000 series, they do not support DDR5 so they must have DDR4 instead. No, you cannot test CPU that support only DDR4 with DDR5. Also in case Intel OFFICIALLY supports higher memory speeds, higher memory speeds are used. We may argue that DDR5 speeds must be exactly same but again, using highest officially supported memory speed is somewhat OK.

Setup is very close to identical except small difference in memory speeds (DDR5) and when DDR5 cannot be used, DDR4 is used instead.

Not that you take some CPU with integrated memory and integrated GPU because you cannot use external GPU or external memory.

Your examples are two software that make very little use of AVX512, you proved it yourself. That is problem. You also use SOC with integrated memory against CPU with external memory. Your comparison is simply invalid because of those two things. I don't even start to list other things that are wrong because those two are more than enough.

OK, we probably CAN estimate M5 performance based on M4 but this was about ARM vs x86-64, not ARM vs ARM?

I asked for standard benchmark procedures. You still cannot give them and considering everything so far, you cannot give them. How about just admitting that at this point there is no proper way to directly compare Apple products vs current x86-64 CPUs? You are trying to do something that simply cannot be done. And refuse to accept that fact.

You need to stop moving the goalposts, I'm tired of proving you wrong even after you move it again again and again.

"OK, we probably CAN estimate M5 performance based on M4 but this was about ARM vs x86-64, not ARM vs ARM?" - if you can't infer the performance of the M5 Max after all I've given you so far then you are just trolling at this point. it's impossible to not have an idea of how it will perform.

"Not that you take some CPU with integrated memory and integrated GPU because you cannot use external GPU or external memory." - nobody cares, I don't care, others here don't care... and for god forsaken reason you care because it gives you an excuse to ignore benchmark results in ALL configurations.

"our examples are two software that make very little use of AVX512" - and yet, the benchmarks and application technical details say otherwise. unlike you I've done my due diligence when researching these things.

"Setup is very close to identical" - aha, so difference memory configs are ok when comparing x86 CPUs, but somehow when adding ARM to the melting pot, the whole internet breaks for you.

"AVX? Both Intel and AMD support AVX, no difference there." - not what you said and asked me to look into. stop moving the goal post.

FYI ARM can run AVX code, you should have guessed that when I showed you blender and other application results. ARM has its own version called Scalable Vector Extension (SVE) and you can use emulation to run non-native code written for x86 AVX. But seriously... did you actually believe the ARM doesn't have native scalable vector support? And just like with AVX, applications have started to add native SVE support.

How well does an x86 CPU emulate SVE?

https://www.ixpug.org/documents/1520629330Jaros-IXPUG-CINECABlender5.pdf

"You also use SOC with integrated memory against CPU with external memory." - you have examples of x86 using integrated memory and benchmark results for those. if you can't accept even those then it all boils down to your argument being "desktop PC doesn't have it so stop comparing them" which makes no sense.

 

[HardReset]

I believe the goalposts moved again…

Exactly, you moved them, there you go:

I want "mythical perfectly equivalent systems" because that is standard benchmark procedure when comparing different CPU architectures.

You need to stop moving the goalposts, I'm tired of proving you wrong even after you move it again again and again.

"OK, we probably CAN estimate M5 performance based on M4 but this was about ARM vs x86-64, not ARM vs ARM?" - if you can't infer the performance of the M5 Max after all I've given you so far then you are just trolling at this point. it's impossible to not have an idea of how it will perform.

First, start telling where I moved goalposts? It seems you both are doing it..

Yeah? We can estimate performance M4 vs M5, agreed. But question was all the time about M4/M5 vs x86-64. Right? If you cannot compare M4 vs x86-64, there is no point comparing M5 vs x86-64 either.

"Not that you take some CPU with integrated memory and integrated GPU because you cannot use external GPU or external memory." - nobody cares, I don't care, others here don't care... and for god forsaken reason you care because it gives you an excuse to ignore benchmark results in ALL configurations.

"our examples are two software that make very little use of AVX512" - and yet, the benchmarks and application technical details say otherwise. unlike you I've done my due diligence when researching these things.

All configurations? There is basically one or two configurations on M4 (and other M4 variants) and gazillion variations on x86-64.

OK, this is Cinebench: https://chipsandcheese.com/p/cinebench-2024-reviewing-the-benchmark

AVX-512 is used in Cinebench 2024, but in such low amounts that it’s irrelevant.

As for Handbrake, Zen 4 vs Zen 5 results clearly indicate AVX512 is used quite little too (as some of performance difference comes from Zen4 vs Zen5). So easily I proven you wrong.

"Setup is very close to identical" - aha, so difference memory configs are ok when comparing x86 CPUs, but somehow when adding ARM to the melting pot, the whole internet breaks for you.

"AVX? Both Intel and AMD support AVX, no difference there." - not what you said and asked me to look into. stop moving the goal post.

Memory config use standard DIMMs, very different from DIMM vs LPDDR5. For some reason, there is no LPDDR5 systems on comparison.

AVX was supported even on Sandy Bridge. AVX512 is supported on Intel Alder Lake and Raptor Lake, for some reason Intel just decided to disable support. If Intel thinks those CPUs are better without it, then it's Intel decision. Not AMD problem definitely.

FYI ARM can run AVX code, you should have guessed that when I showed you blender and other application results. ARM has its own version called Scalable Vector Extension (SVE) and you can use emulation to run non-native code written for x86 AVX. But seriously... did you actually believe the ARM doesn't have native scalable vector support? And just like with AVX, applications have started to add native SVE support.

How well does an x86 CPU emulate SVE?

https://www.ixpug.org/documents/1520629330Jaros-IXPUG-CINECABlender5.pdf

"You also use SOC with integrated memory against CPU with external memory." - you have examples of x86 using integrated memory and benchmark results for those. if you can't accept even those then it all boils down to your argument being "desktop PC doesn't have it so stop comparing them" which makes no sense.

I know SVE. However like with AVX, AVX2, AVX512 etc question is not if it's supported but how Effectively it is supported. Zen5 and Zen5c both support AVX512 but have different effectiveness on support (5c is theoretically about half speed).

Those "benchmarks" sadly include CPUs that Cinebench team says are useless for AVX512 calculations on Cinebench. It is also around 8 years old.

And once again, software that take advantage of AVX512 are shown to have MUCH more performance gain Zen5 vs Zen4 than Blender does. That does indicate Blender does not use AVX512 effectively. Once again, it's not about Support but Effectiveness. Blender seems to be somewhat poor on that Effectiveness. And if, according to your paper, Handbrake AVX512 support is for Skylake era CPUs, it probably sucks.

Those x86 CPUs with integrated memory have very much crippled CPU compared to high performance ones. We are still talking about CPU architecture performance, right? Those CPUs are well behind what architecture is capable for.

 

[Squid Surprise]

Exactly, you moved them, there you go:


First, start telling where I moved goalposts? It seems you both are doing it..

I think the problem is twofold... one, you're not reading the replies you're getting... and two, you're forgetting the nonsense you've posted in your previous posts...

I suggest re-reading all of the posts and realize that the original points you were arguing have changed as you've lost the previous ones... This shouldn't be news to you, as you do it in virtually every thread you post in...

 

[HardReset]

I think the problem is twofold... one, you're not reading the replies you're getting... and two, you're forgetting the nonsense you've posted in your previous posts...

I suggest re-reading all of the posts and realize that the original points you were arguing have changed as you've lost the previous ones... This shouldn't be news to you, as you do it in virtually every thread you post in...

I see you doing this same on basically every time you argue about something. You throw accusations about something that simply does not exist. In this case, you have chance to clearly say what that "nonsense" is, but no. You just say I'm writing "nonsense" without actually pointing out what that is.

As for second paragraph, you have opportunity to state what points I have "lost" but again you fail to do it. That is because you have no idea what you are talking about so you just write some nonsense and hope some readers accept your BS.

What you are doing is trolling at basic level, but sorry, I have too much experience. I put your "posts" here for reference:

lol… you only extol AMD and ignore actual facts… I think we know who the real troll is…

sorry, had a nice laugh at that… could you maybe reread your posts and come back to us? All you’ve done is cherry pick examples where your believed AMD is better…

I see… so you wish to disregard facts unless they agree with your world view… gotcha!

I believe the goalposts moved again…

I think the problem is twofold... one, you're not reading the replies you're getting... and two, you're forgetting the nonsense you've posted in your previous posts...

I suggest re-reading all of the posts and realize that the original points you were arguing have changed as you've lost the previous ones... This shouldn't be news to you, as you do it in virtually every thread you post in...

Basically nothing else than very poor trolling. Better trolls at least try ""

 

[Squid Surprise]

As for second paragraph, you have opportunity to state what points I have "lost" but again you fail to do it. That is because you have no idea what you are talking about so you just write some nonsense and hope some readers accept your BS.

ok, here - let me sum up just about every post you've put on this site: "None of your posts matter, AMD rules!"

happy now?