a look at (ry)Zen's IPC - from Excavator to KabyLake

MrDynamicMan · March 5, 2017

Fine by me. My 4690k is still far more than capable. Also, could we get a graph that doesn't scale by %? It would be more representative.

March 5, 2017

1% and .1% lows are concerning too.

Intel demolishes AMD with consistency of frames at higher frames.

zMeul · March 5, 2017

1 hour ago, tlink said:

hey look, huge performance differences between operating systems. gee maybe we shouldn't completely trample the idea before its even tested.

If this isn't a large performance gap ON THE SAME PROCESSOR than i don't know what is, this shows quite some room for performance improvements if its to be believed. for 8c/16t thats a difference of 26% for min fps, and 8% for the avg fps. thats massive. for 8c/8t its a min fps difference of 8% in min fps, and a 1% difference in avg fps. smaller but the minimum fps difference still is quite significant. hence why there should be more testing of things like this and optimization. like the post says there where other factors influencing it but a difference of 26% is really hard to explain by another factor tbh. not saying im taking this as proof, but more as hints that we shouldn't jump to conclusions too soon.

sir, you are lying to yourself

why? because W7 generates better performance than W10 on any type of CPU, not just Ryzen

and you want to go even further, Linux generates better CPU performance that Windows is general - easily tested with GeekBench

Kobz360 · March 5, 2017

28 minutes ago, zMeul said:

sir, you are lying to yourself

why? because W7 generates better performance than W10 on any type of CPU, not just Ryzen

and you want to go even further, Linux generates better CPU performance that Windows is general - easily tested with GeekBench

are you sure about that my friend?

http://www.techspot.com/review/1042-windows-10-vs-windows-8-vs-windows-7/page6.html

http://www.pcgamer.com/windows-10-but-what-about-performance/

http://www.anandtech.com/show/9543/the-windows-10-review/24

All of those links which, mind you, were published right as Windows 10 launched show a negligible difference, and often better performance on W10 (on these Intel CPUs) between W7, 8.1 and 10

zMeul · March 5, 2017

Just now, Kobz360 said:

are you sure about that my friend?

http://www.techspot.com/review/1042-windows-10-vs-windows-8-vs-windows-7/page6.html

http://www.pcgamer.com/windows-10-but-what-about-performance/

http://www.anandtech.com/show/9543/the-windows-10-review/24

All of those links which, mind you, were published right as Windows 10 launched show a negligible difference, and often better performance on W10 (on these Intel CPUs) between the W7, 8.1 and 10

yes, I am absolutely sure because I did my own testing and what I found was quite different

the only link you quoted there that would be relevant is the techspot one

PCGamer tested at UHD and 1440p? those are GPU bound resolutions

Anandtech shows W8 not W7

Kobz360 · March 5, 2017

18 minutes ago, zMeul said:

yes, I am absolutely sure because I did my own testing and what I found was quite different

the only link you quoted there that would be relevant is the techspot one

PCGamer tested at UHD and 1440p? those are GPU bound resolutions

Anandtech shows W8 not W7

Here's anotha one @ 1080p

https://www.computerbase.de/2015-07/windows-10-spieler-test-amd-nvidia-geschwindigkeit/#abschnitt_benchmarks_windows_10_vs_8_vs_7

If you can provide solid evidence to the contrary, then I'll listen in further.

Kobz360 · March 5, 2017

22 minutes ago, zMeul said:

yes, I am absolutely sure because I did my own testing and what I found was quite different

the only link you quoted there that would be relevant is the techspot one

PCGamer tested at UHD and 1440p? those are GPU bound resolutions

Anandtech shows W8 not W7

and anotha one

http://www.techspot.com/review/1042-windows-10-vs-windows-8-vs-windows-7/page2.html

once again, showing a virtually negligible difference in CPU performance across the last few Windows OS'

Brehohn · March 5, 2017

IPC. Yay. I'm glad my game is above my 75hz refresh rate with Ryzen so overall, I'm happy that I switched from my 4670K.

In my opinion, when reviewers say it's not the best for gaming based on a 20fps difference in the 90's or higher fps area, that's absurd. It's a great gaming chip. If you want the MOST AMOUNT OF FRAMES POSSIBLE FOR ULTIMATE MLG STATUS.... then sure Kaby Lake will offer you those 20 extra frames for no difference in what you see.

zMeul · March 5, 2017

1 hour ago, Kobz360 said:

once again, showing a virtually negligible difference in CPU performance across the last few Windows OS'

since I don't have time to scour through all the test I've done over time, here are some sample runs W7 vs W8.1:

W7:

W8:

consistently better than W8 both in graphics score and physics score

and here is some I did with GTX970 on W7 and W10 insider build:

W7:

W10:

consistently better than W10 in both graphics and physics score

Edited March 5, 2017 by zMeul

79wjd · March 5, 2017

1 hour ago, Brehohn said:

IPC. Yay. I'm glad my game is above my 75hz refresh rate with Ryzen so overall, I'm happy that I switched from my 4670K.

In my opinion, when reviewers say it's not the best for gaming based on a 20fps difference in the 90's or higher fps area, that's absurd. It's a great gaming chip. If you want the MOST AMOUNT OF FRAMES POSSIBLE FOR ULTIMATE MLG STATUS.... then sure Kaby Lake will offer you those 20 extra frames for no difference in what you see.

Yes, but what happens as games get more demanding and it's no longer a question of 70 vs. 90 but 45 vs 60? Ryzen is fine for gaming, but why spend more on an inferior product if your sole intention is gaming? There are only three valid answers to that question.

You're hoping that multithreaded performance overtakes single threaded performance (which at this moment is a complete crapshoot).
You think you may move to a more multithreaded workload.
You're a fanboy.

Sauron · March 5, 2017

21 hours ago, Misanthrope said:

Ooooh the i5 on a platform with motherboards that would be substantially more expensive than the chip itself? That's almost worst.

I didn't know about that, if that rumour is true then it's definitely not a good proposition

Citadelen · March 5, 2017

On ‎03‎/‎03‎/‎2017 at 10:54 PM, Master Disaster said:

-snip-

AMD showed 4K results, so they weren't lying.

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

@djdwosk97 I'd hold my breath on it's true gaming results, first off Windows 10 is having trouble assigning threads/swapping them around, causing massive latency issues in situations that aren't exclusively single threaded or fully maxes all 16 threads. This is why heavily threaded games such as Watch_Dogs 2 gets a performance boost from disabling SMT. Not only that, but the majority of reviewers used the Crosshair VI Hero, which ASUS has now stopped selling due to issues in the BIOS and it bricking itself. Windows 10 also doesn't correctly use the millions of sensors in the CPU properly either, instead of checking every couple of nano seconds I believe it was, it checks every 30 minutes (or 30 seconds I can't exactly remember), this causes issues that mean the CPU doesn't turbo properly and can lead to increased power consumption.

79wjd · March 5, 2017

10 minutes ago, Citadelen said:

AMD showed 4K results, so they weren't lying.

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

@djdwosk97 I'd hold my breath on it's true gaming results, first off Windows 10 is having trouble assigning threads/swapping them around, causing massive latency issues in situations that aren't exclusively single threaded or fully maxes all 16 threads. This is why heavily threaded games such as Watch_Dogs 2 gets a performance boost from disabling SMT. Not only that, but the majority of reviewers used the Crosshair VI Hero, which ASUS has now stopped selling due to issues in the BIOS and it bricking itself. Windows 10 also doesn't correctly use the millions of sensors in the CPU properly either, instead of checking every couple of nano seconds I believe it was, it checks every 30 minutes (or 30 seconds I can't exactly remember), this causes issues that mean the CPU doesn't turbo properly and can lead to increased power consumption.

I've been assuming that ryzen would eventually compete with broadwell-e in gaming (at the same clockspeed). However, kaby lake can still clock much higher and has an IPC advantage. So unless games as a whole benefit more from multithreaded-ness more than single core performance, a 7700k will still be the better choice. The fact still remains that at the end of the day tasks can only be parallelized so much and single core performance will always be important, so saying an 1800x will be better in the long run is just as much of a guess as saying a 7700k will perfom better in the long run. Therefore, the better performing CPU today, for less money, is the smarter choice.

vorticalbox · March 5, 2017

On ‎03‎/‎03‎/‎2017 at 10:26 PM, TOMPPIX said:

why is everyone so fixed on the single core performance, it's multi thread performance destroys every i7 out there in the same price range.

for gaming they are mostly reliant on single core performance but yes in some situations it really shines the thing is the 8350 "really shined" when doing things like compression when compared to an i5.

Brehohn · March 5, 2017

2 hours ago, djdwosk97 said:

Yes, but what happens as games get more demanding and it's no longer a question of 70 vs. 90 but 45 vs 60? Ryzen is fine for gaming, but why spend more on an inferior product if your sole intention is gaming? There are only three valid answers to that question.

You're hoping that multithreaded performance overtakes single threaded performance (which at this moment is a complete crapshoot).

You think you may move to a more multithreaded workload.

You're a fanboy.

I do game only, yes. However, I have a 75hz monitor that I love so could I have bought a $300 i7? Sure. But I wanted something different. I've been with Intel for a LONG ass time. I like having 8 cores and 16 threads under my belt. I'm optimistic that as time moves forward, the updates and optimizations will only get better. Not to mention, DX12 is definitely going to be the future as more adoption becomes available.

79wjd · March 5, 2017

1 minute ago, Brehohn said:

I do game only, yes. However, I have a 75hz monitor that I love so could I have bought a $300 i7? Sure. But I wanted something different. I've been with Intel for a LONG ass time. I like having 8 cores and 16 threads under my belt. I'm optimistic that as time moves forward, the updates and optimizations will only get better. Not to mention, DX12 is definitely going to be the future as more adoption becomes available.

Optimizations and updates to ryzen won't magically make it better than kaby lake. And dx12 won't magically make more (weaker) cores the better answer. If you're willing to gamble that it will then that's something else (and is point number 1). If I personally needed a new cpu I'd definitely consider a 1700 (if/when they overclock better) as a slight performance loss today is worth the risk of a greater performance increase in the future. But that doesn't change the fact that for gaming it is a worse choice.

zMeul · March 5, 2017

29 minutes ago, Brehohn said:

Not to mention, DX12 is definitely going to be the future as more adoption becomes available.

and will do what for Ryzen? that's not already doing for existing DX12 supported tiles

no matter how you turn it, games are IPC dependent than number of available cores dependent - and the fact remains Ryzen's IPC is nothing to write home about

Kobz360 · March 6, 2017

13 hours ago, zMeul said:

since I don't have time to scour through all the test I've done over time, here are some sample runs W7 vs W8.1:

W7:

http://www.3dmark.com/sd/2568810

http://www.3dmark.com/sd/2573782

http://www.3dmark.com/sd/2578046

http://www.3dmark.com/sd/2584312

http://www.3dmark.com/sd/2623617

W8:

http://www.3dmark.com/sd/2657099

http://www.3dmark.com/sd/2657265

http://www.3dmark.com/sd/2657322

consistently better than W8 both in graphics score and physics score

and here is some I did with GTX970 on W7 and W10 insider build:

W7:

http://www.3dmark.com/sd/2827669

http://www.3dmark.com/sd/2828442

http://www.3dmark.com/sd/2848156

http://www.3dmark.com/sd/2848165

http://www.3dmark.com/sd/4155092

W10:

http://www.3dmark.com/sd/3026846

http://www.3dmark.com/sd/3044751

consistently better than W10 in both graphics and physics score

Mr muel to be completely honest, I've provided evidence from multiple reputable sources which backup my claim (and support the general consensus) that performance between Windows 7 and Windows 10 is virtually the same with it often being better on the latter. The tests I provided from these sources would presumably be done so in a way to truly test the performance difference across OS'.

While your first hand data does show the contrary, it's not really enough data or testing at all to really backup your claim that Windows 10 performs worse on any CPU, just as you said so yourself (not to mention you're using W10 insider build and It's just a few 3dmark scores vs. various benchmarks I've provided.). That simply isn't true. It may be the case in some circumstances, as we can see here with Ryzen (and will probably be fixed) and even so with your Core 2 Quad, but it is clearly not the case across the board.

I'm not saying Windows 10 performs much better or even noticeably better than 7, I'm simply saying there's no evidence to support your claim of Windows 10 consistently performing worse on any given CPU, because I've clearly shown that it's not true.

zMeul · March 6, 2017

5 hours ago, Kobz360 said:

I've provided evidence from multiple reputable sources which backup my claim

that's not your claim, you are lead to believe something - do your own testing like I did my own

Kobz360 · March 6, 2017

1 minute ago, zMeul said:

that's not your claim, you are lead to believe something - do your own testing like I did my own

I am lead to believe it because it's been backed up by multiple reputable sources (as I've linked previously). Once again I'm only disputing your notion that Windows 7 performs than 10 on any CPU. I'm not even disputing the fact that you specifically got a regression in performance on W7 vs W10.

zMeul · March 6, 2017

15 minutes ago, Kobz360 said:

I am lead to believe it because it's been backed up by multiple reputable sources (as I've linked previously). Once again I'm only disputing your notion that Windows 7 performs than 10 on any CPU. I'm not even disputing the fact that you specifically got a regression in performance on W7 vs W10.

it's not only me

zMeul · March 8, 2017

Puget Systems took a look at Ryzen: https://www.pugetsystems.com/labs/articles/Premiere-Pro-CC-2017-AMD-Ryzen-7-1700X-1800X-Performance-909/

their conclusion is this:

Quote

Overall, Ryzen is in an interesting option for a Premiere Pro workstation. It is about 5-10% faster than the Core i7 7700K for exporting and rendering previews - although the i7 7700K has a sizable advantage when it comes to tasks like warp stabilize. Since the 1700X and 1800X are about 20-30% more expensive than the i7 7700K, this means that if you are on a budget you have a bit of a choice between the i7 7700K and the new Ryzen CPUs. One thing to keep in mind, however, is that in a number of lightly threaded applications like Photoshop and Lightroom we found that Ryzen is quite a bit slower than the Core i7 7700K. So the choice is to pay 20-30% more for 5-10% faster export and preview generation times, but much lower performance for warp stabilize and in other applications like Photoshop.

Compared to the 6-8 core "High End" Intel CPUs, the two Ryzen CPUs we tested fared pretty well. They should be a bit faster when exporting and rendering previews than the Core i7 6850K (for about $100-200 cheaper as well), but are about 5-10% slower for warp stabilize. If you use warp stabilize in your projects, we would really consider the Ryzen CPUs to be on equal footing with the i7 6850K. Compared to the Intel Core i7 6900K (which is about $500-600 more expensive), the 1700X should be around 7-10% slower while the 1800X should only be around 2-4% slower. So from a price/performance standpoint the 1800X is clearly a better value than the i7 6900K but the i7 6900K is still faster overall.

However, our take on Ryzen for Premiere Pro is that while it presents some interesting opportunities and can be competitive with Intel, it doesn't quite earn a place on our Premiere Pro recommended systems. This isn't to say that a Ryzen workstation might not be the right fit for you, but there are a couple of reasons that (right now) it just barely misses the mark for us:

If you are on a tight budget, the Intel Core i7 7700K is likely a better choice for most users. While the 7700K is a little slower than Ryzen for some tasks in Premiere Pro, but it is much faster for other tasks (like warp stabilize) and in applications like Photoshop. The i7 7700K is a bit cheaper as well which frees up some of your budget for more RAM, larger storage, etc.

For higher-end workstations, the 64GB RAM limitation of the Ryzen platform is a concern. 64GB is becoming much more commonplace in our Premiere Pro workstations as 4K and VR projects are gaining steam and not having the room for future RAM expansion is not ideal.

While these compromises keep Ryzen from earning a spot in our Premiere Pro configurations, this is certainly a big step forward for AMD and we will be watching them much more closely in the future. It is exciting to have more competition and options, and for the first time in a long time, we can imagine a future where AMD will be back in our product line - even if that day isn't today.

79wjd · March 8, 2017

-nvm-

baconbuilder99 · March 8, 2017

might I add that in REAL WORLD testing (not benchmarks) that the issues were at 1080p, but when cranked up to 1440p and 4k, It was right up there with broadwell-e, and is it reellyy worth paying twice as much for X99?

from a value for money and workstation perspective, the r7s start to make a lot of sense

zMeul · March 9, 2017

7 hours ago, baconbuilder99 said:

might I add that in REAL WORLD testing (not benchmarks) that the issues were at 1080p, but when cranked up to 1440p and 4k, It was right up there with broadwell-e, and is it reellyy worth paying twice as much for X99?

from a value for money and workstation perspective, the r7s start to make a lot of sense

BS logic, if you look at those benches, i3s and Pentiums perform similarly ... so why pay for Ryzen?!

Sign In

a look at (ry)Zen's IPC - from Excavator to KabyLake

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

*Insert Name* R̶y̶z̶e̶n̶ Intel Build! https://linustechtips.com/main/topic/748542-insert-name-r̶y̶z̶e̶n̶-intel-build/

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

*Insert Name* R̶y̶z̶e̶n̶ Intel Build! https://linustechtips.com/main/topic/748542-insert-name-r̶y̶z̶e̶n̶-intel-build/

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Link to comment

Share on other sites

Link to post

Share on other sites

Insert Name R̶y̶z̶e̶n̶ Intel Build! https://linustechtips.com/main/topic/748542-insert-name-r̶y̶z̶e̶n̶-intel-build/

Insert Name R̶y̶z̶e̶n̶ Intel Build! https://linustechtips.com/main/topic/748542-insert-name-r̶y̶z̶e̶n̶-intel-build/