i like foxes
Posts: 312 +218
The larger cache and faster the better , it improves IPC
AMD Ryzen 9 9950X Review: We've Seen This Before...
- Thread starter Steve
- Start date
takaozo
Posts: 1,084 +1,665
Thanks!
vicsrealms
Posts: 38 +51
Alright AMD, lets see the X3D versions.
What's wrong with them is that the 7000 series offers almost the same performance for significantly cheaper. It's especially bad if you are after a gaming system because the 7800x3D is both cheaper and faster than any of the 9000 series.
I don't like it because none of the software I use benefits much from the improvements AMD made. As far as I know, there aren't any alternatives that do either. They clearly focused on enterprise performance this generation and that sucks for anyone who uses consumer software. I'm not running a Linux server, and neither are 95+% of people buying AMD's consumer products in the 9000 series.
unoficialoficial
Posts: 204 +155
I think AMD has done a great job in designing the processor architecture despite the constraints of the laws of physics and the nature of the processes involved in improving CPU IPC. I know some people were hoping for more impressive increases in processor performance, but I think the improvements they've made are really good!
First of all, they only had a budget of 1 extra nanometer in the same extreme ultraviolet spectrum (the Ryzen 7000 had an advantage of 2 extra nanometers and a transition from deep ultraviolet to extreme ultraviolet spectrum over the Ryzen 5000).
And that's where it gets really interesting! Almost all ways to improve IPC involve improving predictions. But, as we all know, there are only so many ways to improve predictions. At some point, it just reaches the upper limit. For instance, because there's a practical limit to predicting how a program's instructions will evolve, it's not possible to achieve a proportional increase in performance by using more cores (Amdahl's law). It's a well-known fact that doubling the number of processing cores doesn't necessarily mean doubling the performance, except for a few rare cases where it's possible to achieve 100% logical segmentation.
I also feel that AMD may have overvolted this chip a little too much (to be sure that it's stable) but on its 4nm super dense transistors package under certain conditions that extra voltage can cause chain reactions and destabilize the whole operation. I don't think there's ever been a case in the history of CPUs where a new processor from the same company has underperformed in any kind of process. It's hard to believe that the same architecture at the same GHz with all those improvements in instruction scheduling, branch prediction, speculative execution, instruction window, cache, data prefetching, micro op fusion, power management, etc. could possibly perform worse on 7-zip compression and uncompression and even with higher power consumption. This just doesn't seem possible under normal conditions, so I'm wondering if there might have been a mistake somewhere.
It will be exciting to see what the future holds for us in terms of new materials with lower electrical resistance. With IPC improvements through more accurate predictions reaching a point of diminishing returns, it's clear that the only other way to improve processor performance is to increase clock speed via the use of new materials.
First of all, they only had a budget of 1 extra nanometer in the same extreme ultraviolet spectrum (the Ryzen 7000 had an advantage of 2 extra nanometers and a transition from deep ultraviolet to extreme ultraviolet spectrum over the Ryzen 5000).
And that's where it gets really interesting! Almost all ways to improve IPC involve improving predictions. But, as we all know, there are only so many ways to improve predictions. At some point, it just reaches the upper limit. For instance, because there's a practical limit to predicting how a program's instructions will evolve, it's not possible to achieve a proportional increase in performance by using more cores (Amdahl's law). It's a well-known fact that doubling the number of processing cores doesn't necessarily mean doubling the performance, except for a few rare cases where it's possible to achieve 100% logical segmentation.
I also feel that AMD may have overvolted this chip a little too much (to be sure that it's stable) but on its 4nm super dense transistors package under certain conditions that extra voltage can cause chain reactions and destabilize the whole operation. I don't think there's ever been a case in the history of CPUs where a new processor from the same company has underperformed in any kind of process. It's hard to believe that the same architecture at the same GHz with all those improvements in instruction scheduling, branch prediction, speculative execution, instruction window, cache, data prefetching, micro op fusion, power management, etc. could possibly perform worse on 7-zip compression and uncompression and even with higher power consumption. This just doesn't seem possible under normal conditions, so I'm wondering if there might have been a mistake somewhere.
It will be exciting to see what the future holds for us in terms of new materials with lower electrical resistance. With IPC improvements through more accurate predictions reaching a point of diminishing returns, it's clear that the only other way to improve processor performance is to increase clock speed via the use of new materials.
amghwk
Posts: 1,707 +1,719
Despite Steven Walton's smear campaign against AMD CPUs this round, there's no denying that these Zen 5 CPUs are extremely powerful single threaded and multi threaded workhorse. The price? Yeah, it's high, but nobody is waiting to upgrade right now. When the price comes down later, these CPUs will offer much better value.
(This is coming from a user who has been alternating between Intel and AMD CPUs while upgrading.)
(This is coming from a user who has been alternating between Intel and AMD CPUs while upgrading.)
Mr Majestyk
Posts: 2,423 +2,256
Please note Phoronix mostly tested AVX-512 enabled apps. They even said stay tuned for non-AVX-512 tests soon. We already knew AVX-512 can make a huge difference like in y-Cruncher even on Winblows."The Ryzen 9 9950X was 33% faster than the Intel Core i9 14900K performance overall and even the Ryzen 9 9900X was 18% faster than the Core i9 14900K. For those still on AM4, the Ryzen 9 9950X was delivering 1.87x the performance of the Ryzen 9 5950X processor. These are some great gains found with the Ryzen 9 9900 series."
-Phoronix
Then you don't understand how to test CPU gaming performance.
Sounds like coping based on wishes.
And I say that as someone that has 3 AMD desktops currently (3000 & 5000 series). I skipped the 7000 series because of overpriced RAM (at the time) but the 9000 is such a disappointment, I’m just going to get a 7800x3D to hold me over to the 11000 series.
Then use better software. Sad truth is that improving x86/SSEX performance gives very small double digit performance gains. With AVX and especially AVX512 50-100% performance gains are realistic.
Just like with Pentium 4 SSE2 and Athlon64 X86-64, performance gains were minimal until software catched up. But would you really have wanted to stick with 32-bit software without SSE-support until now?
Accept price of going forward or stagnate forever. That's how it is.
Soulburn74
Posts: 150 +80
Being a 4K gamer mostly, and a little bit of productivity (a little bit of Premiere and Photoshop work), with a 3950x and a 4090: There hasn't really been much to persuade me into moving up to a 5950x/7950x, and now with results of the 9950x coming in that shows not much improvement for my mostly 4k gaming perspective. I'll stick with what I have...
My biggest gripe, not really related to this article, (and I realize these days I'm in a niche scenario), is the lack of pcie lanes..... Without going Threadripper its basically impossible to have x16/x16 mode in two pcie slots on current motherboards. I'm sure the use case for this is very small these days. Most people have moved onto software raid solutions like unraid etc, vs using a sas controller. But I'm one of those few people.....and I'd be nice not to have to downgrade my gpu performance in order to also have a SAS/SATA controller in my rig......
My biggest gripe, not really related to this article, (and I realize these days I'm in a niche scenario), is the lack of pcie lanes..... Without going Threadripper its basically impossible to have x16/x16 mode in two pcie slots on current motherboards. I'm sure the use case for this is very small these days. Most people have moved onto software raid solutions like unraid etc, vs using a sas controller. But I'm one of those few people.....and I'd be nice not to have to downgrade my gpu performance in order to also have a SAS/SATA controller in my rig......
Well, the normal AMD shills aside, this is kind of sad... I guess the result of being CPU kings for the past several years have turned AMD into the old Intel... resting on their laurels and giving incremental improvements only... Still, at least it uses the same motherboards as the last generation, so you can upgrade an older 7000 series in a year or 2 for a bargain - assuming these CPUs come down in price.
I remember people telling me I was a fool to buy a 7980x Threadripper back in the winter as the 9000 series would blow it out of the water at a fraction of the price... well, glad I ignored that BS...
I remember people telling me I was a fool to buy a 7980x Threadripper back in the winter as the 9000 series would blow it out of the water at a fraction of the price... well, glad I ignored that BS...
Weird that Steve and TS have always advocated that Ryzen x950X processors are not meant for gaming then bash 9950X for being a bad gaming cpu. Wild times.
These results prove AMD really delivered: https://openbenchmarking.org/vs/Processor/AMD+Ryzen+9+9950X+16-Core,AMD+Ryzen+9+7950X+16-Core
If people want to use outdated software, that's not AMDs fault.
I have already posted many links that prove 12+ percent improvement. This Zen5 sucks agenda is like complaining how AMD is slow on Pifast
Nobina
Posts: 4,651 +5,756
So, are these results on TS false?I have already posted many links that prove 12+ percent improvement. This Zen5 sucks agenda is like complaining how AMD is slow on Pifast
Burty117
Posts: 5,608 +4,488
The problem with that arguement is that my favourite software (games mainly) already exist and have close to zero benefit from the new processors.
What do you want me to do? Email the Helldivers 2 developers...
"Ever so sorry, HardReset on the internet told me your software is legacy, can you completely re-write your game please using specifically the AVX512 instruction set? Okay Thanks Bye!"
My point being, AMD exclusively concentrating on "modern" instruction sets is nice, but practically no use to most gamers. I guess PS3 emulation must be pretty tight on the new CPU's though.
As I said, that software doesn't exist. No one should buy a new CPU hoping that someday it might be faster in future software. I'd be especially cautious in this case since AVX512 has been around for a while and it hasn't seen wide adoption in consumer or even professional software.
I'm not saying they are false, since performance depends on used software, hardware, settings...
However, it has been proven multiple times that Ryzen performance on certain workloads is excellent. That means Zen5 is not small but very big improvement from Zen4. If that does not show on TS workloads, it still doesn't mean Zen5 is "flop".
That your POV. However, if you want advancement, you also have to accept sacrifices. Good example:
x86-64 compilers make only SSE2 or higher FPU code, not x87. You can still use x87, but compilers do not make x87 code. So, if 64-bit software is used, why bother with x87 performance at all? Well, someone wants Pifast to run much faster but price is either hgher CPU price or lower performance on other areas. Choose:
- Superior performance in Pifast, more expensive CPU OR less performance onother areas
- Worse performance on Pifast but better performance on 64-bit software AND no additional CPU cost
You cannot please everyone.
Similar threads
Latest posts
-
AMD Ryzen CPUs continue to crush Intel processors on Amazon best-seller list- Daniel Sims replied
-
Study shows gaming can reduce stress, even the violent kind- captaincranky replied
-
Philips Hue confirms upcoming US price hikes are due to Trump tariffs- GodisanAtheist replied