AMD Ryzen Gaming Performance In-Depth: 16 Games Played at 1080p & 1440p

Thanks for the review. I have been waiting to build a new computer, casual gaming, photography, etc. I think for the first time I will be building an AMD build. I am waiting for their new video cards to be released before I move away from nvidia. I will be buying everything but the video card in may. I love your site.
 
The 7700K beats the crap out of Intel's $1700 6950X. Clearly you have no idea CPUs are primarily used for other tasks that AREN'T gaming, and are priced off THOSE benchmarks. In productivity AMD is basically equal to Intel's offerings at 1/2 the price. IT professionals/Businesses probably can't wait to get their hands on these things considering the major cost savings. AMD will make 100x the money off that segment rather than the gaming department. Regardless the 6-8 Core AMD or Intel CPUs are still better long term solutions vs any quad core. Unless you like upgrading your entire system every year or two.
very well said sir, this article confirms what you are telling

https://www.servethehome.com/amd-ryzen-7-1700x-linux-benchmarks/
 
Did TechSpot tell the readers that they had disabled Asynchronous Compute?

NVidia does not support Asycnh Compute and that is a major AMD performance multiplier while running DX12 and Vulkan games.

In fact while RX 480 may not have the performance of say GTX 1080, 2x RX 480 is also FULLY supported in DX12 and Vulkan and 2xRX 480 outperforms GTX 1080 for far less money.

This was a LIE OF OMISSION.

DISAGREE???

Then run the benchmarks with Async Compute ENABLED with 2x RX 480. Your becnhmarks here crippled Ryzen.

You also "FORGOT" to run ChessBase FritzMark.

http://en.chessbase.com/post/amd-releases-new-ryzen-processor


RYZEN SLAMS ALL 8 COTE 16 THREAD INTEL CPU,s.
 
Nice article Steve! Nicely laid out and plenty of detail, Even though AMD were not at the top of the charts they are still miles better than anything they've produced before. Very impressed AMD, After years of let down, you've come out swinging this time.

I wonder when we'll see a patch (presumably from Microsoft) that'll fix the whole issue of Windows guessing the caches wrong on Ryzen (Windows is counting each thread as a full core). I bet this will help sort the SMT issues at least since Windows will be able to tell which threads are actual cores.

I am not sure that SMT is a Windows problem.

Chessbase FritzMark is a non-GPU dependent benchmark and it has no problem at all giving good results.

http://en.chessbase.com/post/amd-releases-new-ryzen-processor

I think the issue is far more than a Windows problem. In fact I think it is a basic incompatibility with nVidis while running thre dGPU with Asynchronous Compute disabled.

FrtizMark recognizes up to 64 cores and is a very sophisticated system stress test and of course it runs on Windows.
 
No future mensa candidate because what you linked is a techspot report showing faster memory displaying an improvement in FPS which is both not what the OP described in his post or what is even up for debate as most people agree to that claim. What I am talking about is this article stating that while 16 GB of RAM is desirable it's not needed compared to 8GB and shows little improvement. Therefore, a system set up just for game testing would not show any difference between 16 GB and 32 GB.

Those building a new system or simply looking to upgrade their memory capacity, the answer is simple: 8GB should be the minimum standard, while 16GB is desirable but not needed. For general usage and gaming there is no advantage to be had by using 16GB or more RAM, though admittedly system memory has hit new lows in 2016 which makes it very accessible even in budget builds.


In gaming scenarios we were surprised to see 4GB will help you extract most out of your system (actual gameplay frame rates versus say, loading levels) but 8GB remains ideal. Those of you focused solely on gaming who don't have the extra cash to splurge on 16GB of memory, fear not, you aren't missing out on any hidden performance. For folks who insist that certain game mods will use over 8GB of RAM, that's fine, go for it, but as far as we're able to determine there aren't any popular games that require over 8GB without mods.

https://www.techspot.com/article/1043-8gb-vs-16gb-ram/page4.html

There's no difference between 8GB/16GB/nor 32GB... didn't think that question would even come up.
 
Pity is, that the price for 3600MHz RAMs is much higher than 3200MHz or 2667Mhz, and thats just because AMD dont decided for a quadchannel RAM

Going quad channel memory would have increased the cost of the processors and the motherboards and not all workloads benefit from quad ram so it was a trade off they had to make.
 
At $500 for R7 1800x, AMD = Another Marketing Deception. Who is willing to be the gullible marks to be conned? They need to price their stuff according to the these benches. The R7 1800x at best can be price equal to the i7-7700k, and go down from there. As consumers we do NOT have the drink their Koolaid about how they are competing against the 6900K or what not. We should demand the best value for our dollar.

Using that logic, i7-6900K should be priced lower than i7-7700K. Good luck explaining why.

He can't because these businesses are not running a charity.

They will price it at what the market will bare.
 
I'm still impressed with Ryzen. They're very competitive in both price, and at 1440p, they close the gaming performance gap. Now I just have to decide if I'll do the 1700X or 1800X.
 
There is more into gaming I wish but nobody ever does. What about multibox. An example is archeage which allows multiple alts in order to transfer packs. One alt = one account/window. So many people including me, run 2-3-6-10 windows at the same time! My old i7 870 can run only 2 windows. So I suspect that when it comes into multibox, RYZEN is the way to go... many cores = many windows.

Another good scenario for benchmark is alpha star citizen, you know how many want to play it but its tooooo heavy at this point. My i7 870 cannot play it in the open mini universe they got.
 
Overall I would love to see how ryzen against intel performs with these 2 games and in multibox. It is economy sandbox games and now most mmos turn into that, economy driver pvp/sandbox mmos. Other games to follow up the same philosophy are chronicles of elyria, etc.
 
Just not quite good enough. In some games even the 1800X loses by a pretty enormous amount to the faster Intel quad processors. This is also in 1440p where the game leans heavier on the GPU. In Watch Dogs 2 7700K is 50 percent faster. In Far Cry Primal and Warhammer the 7700K is more than 35 percent faster minimum frame rate. In Arma 3 and GTA5 it's 25 percent.

These are not small games and the results are pretty brutal when you actually read the figures like this.

The results look even more problematic at 1080p, where the vast majority of mainstream gamers still play at and will be looking at the mainstream parts like the 4 cores.

So the gaming performance is questionable at 1440p on 8 cores with all these threads and cache, the real issue is how much slower are the mainstream stripped down 4 and 6 cores going to be at 1080p?

Ryzen needs highly clocked quads to compete in the mainstream gaming space. I have my doubts whether AMD can deliver the sort of clock speeds and boost speeds required to compete against the better Kaby Lake quads. The fact that the 1800X struggles to get beyond 4ghz overclocked without extreme measures while a Broadwell 8 core can pretty much cruise along at least 400-500mhz faster than that suggests Intel's architecture is considerably more refined.

Ryzen may need a bunch of hardware revisions before I'm optimistic about the clock speeds of the cheaper parts.

I will admit that these 8 cores aren't really gaming parts, but their performance does not bode well right now for the cheaper ones. If you are building a machine you are wanting to game a bunch on you still have to go Kaby Lake.
 
Last edited:
Did TechSpot tell the readers that they had disabled Asynchronous Compute?

NVidia does not support Asycnh Compute and that is a major AMD performance multiplier while running DX12 and Vulkan games.

In fact while RX 480 may not have the performance of say GTX 1080, 2x RX 480 is also FULLY supported in DX12 and Vulkan and 2xRX 480 outperforms GTX 1080 for far less money.

This was a LIE OF OMISSION.

DISAGREE???

Then run the benchmarks with Async Compute ENABLED with 2x RX 480. Your becnhmarks here crippled Ryzen.

You also "FORGOT" to run ChessBase FritzMark.

http://en.chessbase.com/post/amd-releases-new-ryzen-processor


RYZEN SLAMS ALL 8 COTE 16 THREAD INTEL CPU,s.

giphy.gif
 
I couldn't find a single review that showed it beating a 4790k in gaming?[/QUOTE]
Yes, at 20% lower clocks it outperforms the fail-lakes by 5%+ in CPU bottlenecked gaming situations while using less power - saying it's "slower and lower powered" is quite a misnomer given it OC's over 4GHz with ease. BTW, it's merely a well binned and higher clocked i7-5775c. Don't take my word for it:
http://www.anandtech.com/show/9532/...on-e3-v4-review-95w-65w-35w-1285-1285l-1265/7
http://techreport.com/review/28751/intel-core-i7-6700k-skylake-processor-reviewed/6
 
My problem is not even the gaming performance, because that could be improved with patches.
I Have mainly 3 Problems with the Ryzen(and im an AMD fun):

1. Why Dual Channel RAM.... At the moment as I saw even if the mainboards can partially OC the RAMs even to 3600 MHz, but just on 2 slots, if I use 4 slots, the speed would drastically fall. I hope thats possible to change with a bios update

2. Overclocking and XFR - was so much wrote about the XFR..and even if I push a Predator 240 or Predator 360 I can get +100 Mhz from XFR.. that very very disapointing, the same about overclocking...the 6900 is able to overclock to 4,4Ghz and by the Ryzen 1800X is by 4,1 the end.....by the XFR range...

3. 20 PCI Lanes. Why to make X370 chipset for SLI setups with 20 PCI lanes. SLI setup is mostly made with high end GPUs which need x16 for the best performance. 20 Lanes are definetly not enough for a good modern processor. there should be at least 40. 2x 16 for the SLI setups and 2x 4 for two M2 or U2 slots.
1. Intel uses quad channel ram on Xeons with 22 cores, why does an 8 core need more than dual? x99 chips show negligible performance lose on dual channel for most applications. You have a point and the 4 dimm stability.

2. I don't know enough about silicone lithography to be for sure about this, but it comes down to a few things. AMD does not control the process, so this 14nm process was probably originally made with mobile chips in mind, with a focus on efficiency over out right performance.

Density, according to public numbers Ryzen is 25 mil transistors per a square mm and Intel's current chips all workout to about 15 mil per mm². High density should help prices with more chips per a wafer but hurt the chips ability to clock higher.

I'm not expert so that is all speculation.

3. More PCI lanes is more cost. For mainstream applications no GPU released yet will experience more than a few percent or less hit to performance running in pci-e 3 x8 vs x16. This has been tested and verified.

For the m.2 drives, if you want to run 2 x4 m.2 drives on a comparable price Intel platform, you are going to be running both drives with x4 pci-e connections to a PCH that is connected to the processor by... Pci-e 3 x4 speeds or stealing from the 16 direct lanes for gpu.

For the price brackets this platform and cpu's fall into, I feel this is all very fair. We even get bonuses like ecc memory being allowed. I don't know why ecc memory is not the only memory at this point, but where AMD does not pay to certify ecc support like servers need they still allow it. Intel artificially blocks ecc support encourage you to spend more money.

Instead of burning Ryzen for its faults, let's identify them and hold AMD accountable for fixing what it can. Let's buy Ryzen if it makes sense for our use case. And let's enjoy the first competition among x86 CPUs in years.
 
Nice article Steve! Nicely laid out and plenty of detail, Even though AMD were not at the top of the charts they are still miles better than anything they've produced before. Very impressed AMD, After years of let down, you've come out swinging this time.

I wonder when we'll see a patch (presumably from Microsoft) that'll fix the whole issue of Windows guessing the caches wrong on Ryzen (Windows is counting each thread as a full core). I bet this will help sort the SMT issues at least since Windows will be able to tell which threads are actual cores.

I am not sure that SMT is a Windows problem.

Chessbase FritzMark is a non-GPU dependent benchmark and it has no problem at all giving good results.

http://en.chessbase.com/post/amd-releases-new-ryzen-processor

I think the issue is far more than a Windows problem. In fact I think it is a basic incompatibility with nVidis while running thre dGPU with Asynchronous Compute disabled.

FrtizMark recognizes up to 64 cores and is a very sophisticated system stress test and of course it runs on Windows.

You had me at "I'm not sure".
 
Nice article Steve! Nicely laid out and plenty of detail, Even though AMD were not at the top of the charts they are still miles better than anything they've produced before. Very impressed AMD, After years of let down, you've come out swinging this time.

I wonder when we'll see a patch (presumably from Microsoft) that'll fix the whole issue of Windows guessing the caches wrong on Ryzen (Windows is counting each thread as a full core). I bet this will help sort the SMT issues at least since Windows will be able to tell which threads are actual cores.
I am curious to see as well what software patches for software like Windows itself will do for Ryzen over the next few months. Hopefully the 4 and 6 core Ryzens are not a let down.
 
I am just wondering if ryzen cpu + upgraded gpu (better than gtx 960) can have a massive improvement over my i3-3240 + gtx 960 in so far as handbrake performance is concerned. I can encode x265 video from retail dvd in 2h+. if ryzen can do it an hour less, I will seriously consider it. already got 350$ but this is for my next gpu. still to save 500$+ equivalent for the possible ryzen + x370 am4 motherboard + ddr4.
 
Can some one tell me why Chess Games are NOT included while testing CPU for its performance. Why only those games where ONLY Frame Rates are important ?
at least nobody said: yeah! but can it run crisis better? ;)
 
Some will look at the graphs and tell they are disappointed, but one must consider the price. For 30-90% less money you get the same performance as Intel. People that want the very best in gaming will still spend $800-1000 on Intel, while people that want best bang for the buck should take Ryzen no questions asked.

What? No, no one should ever buy an i7 6900K for gaming, it's a rip-off. The king of gaming right now is the 7700K which is $330, which is cheaper than both the 1800X and the 1700X and it outperforms both of them. So for strictly gaming I would not buy an 1800X, a 1700X, an i7 6900K, nor an i7 5960X.
 
Can some one tell me why Chess Games are NOT included while testing CPU for its performance. Why only those games where ONLY Frame Rates are important ?

Because it's easier to get the FPS directly from a program, than trying to get the time difference in milliseconds... either keeping time manually.... or who knows how to get the exact time for CPU "thinking" period using a program, if it exists....
 
My problem is not even the gaming performance, because that could be improved with patches.
I Have mainly 3 Problems with the Ryzen(and im an AMD fun):

1. Why Dual Channel RAM.... At the moment as I saw even if the mainboards can partially OC the RAMs even to 3600 MHz, but just on 2 slots, if I use 4 slots, the speed would drastically fall. I hope thats possible to change with a bios update

2. Overclocking and XFR - was so much wrote about the XFR..and even if I push a Predator 240 or Predator 360 I can get +100 Mhz from XFR.. that very very disapointing, the same about overclocking...the 6900 is able to overclock to 4,4Ghz and by the Ryzen 1800X is by 4,1 the end.....by the XFR range...

3. 20 PCI Lanes. Why to make X370 chipset for SLI setups with 20 PCI lanes. SLI setup is mostly made with high end GPUs which need x16 for the best performance. 20 Lanes are definetly not enough for a good modern processor. there should be at least 40. 2x 16 for the SLI setups and 2x 4 for two M2 or U2 slots.
1. Intel uses quad channel ram on Xeons with 22 cores, why does an 8 core need more than dual? x99 chips show negligible performance lose on dual channel for most applications. You have a point and the 4 dimm stability.

2. I don't know enough about silicone lithography to be for sure about this, but it comes down to a few things. AMD does not control the process, so this 14nm process was probably originally made with mobile chips in mind, with a focus on efficiency over out right performance.

Density, according to public numbers Ryzen is 25 mil transistors per a square mm and Intel's current chips all workout to about 15 mil per mm². High density should help prices with more chips per a wafer but hurt the chips ability to clock higher.

I'm not expert so that is all speculation.

3. More PCI lanes is more cost. For mainstream applications no GPU released yet will experience more than a few percent or less hit to performance running in pci-e 3 x8 vs x16. This has been tested and verified.

For the m.2 drives, if you want to run 2 x4 m.2 drives on a comparable price Intel platform, you are going to be running both drives with x4 pci-e connections to a PCH that is connected to the processor by... Pci-e 3 x4 speeds or stealing from the 16 direct lanes for gpu.

For the price brackets this platform and cpu's fall into, I feel this is all very fair. We even get bonuses like ecc memory being allowed. I don't know why ecc memory is not the only memory at this point, but where AMD does not pay to certify ecc support like servers need they still allow it. Intel artificially blocks ecc support encourage you to spend more money.

Instead of burning Ryzen for its faults, let's identify them and hold AMD accountable for fixing what it can. Let's buy Ryzen if it makes sense for our use case. And let's enjoy the first competition among x86 CPUs in years.

I can full accept your arguments, but than they should say, we made a medicore processor for the price. And not tell everywhere we made a competitive processor for the i7-6900k......because that simply not true. Yes in 1 point they made...The raw arythmetic performance is same or even higher, but if the other things are worst(memory bandwith, less PCI-lanes, even in SSD Overall performance is much weaker) than I cant tell is the same performance for less money.

Its like I build a Ferrari like motor, but put in a small car, maybe is able to go even faster in some cases, because less weihght, but in RL u cant use the whole car, because the brakes and everything else is much weaker and hold back.

Still the Ryzen is not bad for the money
 
It's a shame to see the 7700k recommended for high Hz gaming as Broadwell DT (5775C and e3 1285v4) absolutely destroy it. That said, concluding the R7 1700 is a great buy definitely hits the mark. Now if only AM4 motherboards were decent - especially in the SFF realm...

How does a 2 years old e3 1285 v4 destroy a 7700k in gaming? Link me please.
 
(Commented elsewhere because comments were initially not available here, but thought it's worth repeating.)

Would be cool to see 4 cores (4+0), not SMT tests against Core i5's (and possible i7's) of various generations.
 
Back