GeForce GTX 1080 Ti: Core i7 7700K vs. Ryzen 7 1800X Linux Gaming Performance

[leipero]

Some guys inspected AdoredTV's claim how BD is now faster than SB and found that opposite is actually true...

On average nothing happened, same true like 5 years ago

"Some guys" isn't really a scientific claim..., AdoredTV used benchmarks from other websites (not his own), and he didn't claimed that Bulldozer is "now faster" than SB i5, it always was faster, still is, and always will be, it's that simple. He claimed that games used max 2 cores when BD came out, and now, games that can use more than 4 cores (very few) are actually faster on BD, and that's only logical since 8 cores (even with shared FPU) > 4 cores.

Anway, can you link me to that "debunking" video/article of AdoredTV?

[leipero]

Optimizations of what?

Ryzen is simply an inferior architecture in terms of gaming performance.

I don't see how superscalar is connected with this at all, being superscalar just means doing x instruction sets per cycle, wth does that have to do with anything? As far as I know, all the workloads (except workloads where is lack of instruction sets) work on pretty much same level, except the fact that Ryzen have quite lower latency for threads than KabyLake CPU's (so, multi threading will be/is more efficient on Ryzen, independent of workload).

So, if you measure FPU workload (likce Cinebench) you can clearly see that Ryzen is at level of KL or above (given same frequency), and as far as I know, there's no magical "gaming performance" for CPU, games are heavily dependent on floating point workloads (on FPU in CPU).

So you made some pretty bold statements here, without going in depth.

[dungeon]

"Some guys" isn't really a scientific claim..., AdoredTV used benchmarks from other websites (not his own), and he didn't claimed that Bulldozer is "now faster" than SB i5, it always was faster, still is, and always will be, it's that simple. He claimed that games used max 2 cores when BD came out, and now, games that can use more than 4 cores (very few) are actually faster on BD, and that's only logical since 8 cores (even with shared FPU) > 4 cores.

Anway, can you link me to that "debunking" video/article of AdoredTV?

Here you go, he exactly looked at article results year over year and claim how now FX getting faster than i5 2500K

At about 10:50 you have chart year after year - 2012 all to the 2017. to prove how AMD's FineWine works in practice

[bug77]

There is no lack of "optimization" for AMD here, this is a simple fact that unoptimized code runs better on Intel's architecture. And it can't be easily fixed either, games would have to be completely rewritten ...

I'm not disagreeing with anything you've said, but the above is the exact argument intel made in favour of Netburst: when compiling for its long pipeline, Netburst could best Thouroughbred/Palomino/etc. And while that was technically, in the end almost nobody saw fit to compile general purpose code for a specific vendor.

That's why I refrain from predicting anything when it comes to Ryzen. It is what it is. Even if it will get better, there's no way for us to tell right now. More to the point, we can't predict how much better it will get. Because as a new platform, it's bound for at least some improvements.

[leipero]

Here you go, he exactly looked at article results year over year and claim how now FX getting faster than i5 2500K

At about 10:50 you have chart year after year - 2012 all to the 2017. to prove how AMD's FineWine works in practice

I think you misunderstood what has been said here, also, you linked his video to me, I've already watched it . He claimed exactly what i wrote tho, what i was interested in, is where are results contradicting those charts? I remind you, this is not his testing, I really don't know if he done any tests whatsoever, or if he even have Ryzen in his hands. He used multiple sources of other people testing it .

[dungeon]

I think you misunderstood what has been said here...

I don't think there was misunderstandings, AdoredTV took random results from some site and interpreted results like they interpreted it and others found that these and other results were mostly wrong, to claim how now FX is faster in year 2017. by +10% than SB.

On Hardware Unboxed's video you can see (at about 15:00 minute) that actually they got +20% results average on 16 games but now SB is faster or in whole 30% difference, which is too much disparity between two stories really

[bakgwailo]

bug77
But bulldozer was never a flop. Sandy Bridge i5's beated FX on regular basis in games till recently, now FX takes the cake, by large margin in any properly multithreaded game. Ryzen is nothing to be compared to bulldozer architecture..., first, bulldozer had 50-60% lower IPC compared to Sandy Bridge, Ryzen have on pair IPC with Kaby Lake. Your comparison is unjust and invalid.

I could write something I'm saying for years now..., but i will wait till BIOS and software fixes come out. But i can predict what will happen, and remmember my words, Ryzen will still be behind Intel CPU's at same frequencuies in games, even tho it beats them in any other task. And if that happens (and I'm 99% sure it will), i will tell you my opinion why (tho, you can assume).

https://www.youtube.com/watch?v=76-8-4qcpPo

The FX-8350 is not faster than sandy bridge, and has not gotten that much faster over the years. I should also know, given I have owned one since release day. Bulldozer was a flop, and Piledriver only saw moderate fixes. Great for some multithreaded workloads, shit IPC/etc. Also, AMD themselves have said IPC is closer to Broadwell-E (not Kaby).

[efikkan]

I don't see how superscalar is connected with this at all, being superscalar just means doing x instruction sets per cycle, wth does that have to do with anything? As far as I know, all the workloads (except workloads where is lack of instruction sets) work on pretty much same level, except the fact that Ryzen have quite lower latency for threads than KabyLake CPU's (so, multi threading will be/is more efficient on Ryzen, independent of workload).

Ryzen is more superscalar than Kaby-Lake (etc.), which is the only thing which can give it an advantage.

So, if you measure FPU workload (likce Cinebench) you can clearly see that Ryzen is at level of KL or above (given same frequency), and as far as I know, there's no magical "gaming performance" for CPU, games are heavily dependent on floating point workloads (on FPU in CPU).

If you read my whole post you'll see that gaming usually is not superscalar, and is generally very inefficient CPU code with loads of branch mispredictions and cache misses. This is also why Bulldozer back in the day sucked at gaming, even though Ryzen is a bit better.

I'm not disagreeing with anything you've said, but the above is the exact argument intel made in favour of Netburst: when compiling for its long pipeline, Netburst could best Thouroughbred/Palomino/etc. And while that was technically, in the end almost nobody saw fit to compile general purpose code for a specific vendor.

Netburst wasn't beaten due to lack of "optimized code", it had terrible penalties for branch mispredictions which caused it to be beaten by a much simpler CPU from AMD.

That's why I refrain from predicting anything when it comes to Ryzen. It is what it is. Even if it will get better, there's no way for us to tell right now. More to the point, we can't predict how much better it will get. Because as a new platform, it's bound for at least some improvements.

Why will it be better? Please elaborate.

[indepe]

A number of optimizations will come very easily, as shown by the fact that several games have a significant performance increase (for example more than 10% for Deus Ex in the test I referenced above) just by disabling SMT. Games should be able to easily gain further by using finer-grained thread control.

Also, my understanding is that DX 11 already has multi-threading, if so then games (or game engines) don't have to be converted to DX 12 in order to improve further on Ryzen 7.

[efikkan]

A number of optimizations will come very easily, as shown by the fact that several games have a significant performance increase (for example more than 10% for Deus Ex in the test I referenced above) just by disabling SMT. Games should be able to easily gain further by using finer-grained thread control.

Synchronization of threads is not the problem.

Also, my understanding is that DX 11 already has multi-threading, if so then games (or game engines) don't have to be converted to DX 12 in order to improve further on Ryzen 7.

"Multithreading" has been supported by all APIs since the late 90s.