AMD THREADRIPPER VS INTEL SKYLAKE X

[Chuk_Chuk]

Found this interesting link over on the intel sub reddit. It talks about how instructions are executed on Ryzen processor.

Seems that in theory Ryzen could have a better IPC than intel in certain workloads. Potentially something that could be changed if AMD could get there own compiler out there.(?)
Also the reason why AMD SMT is better than intels:

This is where simultaneous multithreading comes in. You can run two threads in the same CPU core (this is what Intel calls hyperthreading). Each thread will then get half of the resources. If the CPU core has a higher capacity than a single thread can utilize then it makes sense to run two threads in the same core. The gain in total performance that you get from running two threads per core is much higher in the Ryzen than in Intel processors because of the higher throughput of the AMD core (except for 256-bit vector code).

Recommend reading the whole thing.

 

[zornyan]

I'm not disputing that ^^^^




What? yet more exaggeration, 5Ghz 7700K's with thier lids intact are rare, 5.2Ghz 7700K's are an extremely rare sight, even 8Pack's signature 7700K is 5.1Ghz and that'll be just about the best 7700K you can get.

Oh and love the preprepared disclaimer there, we all know Intel will send you a bunch of pre-binned CPU's to then bin out of and then claim they are the norm

I'm talking from personal experience, I've done about 40 or so 7600k/7700k systems over the last few months and i think only 3 of them couldn't do 5ghz (stuck at 4.9ghz) I've had one do 5.2, but was bought from silicone lottery as a 5.2ghz binned unit (a friend wanted the best he could possibly get) and a few do 5.1

yes, this is a small sample size, but as said I'm merely talking from my experience and that of people I know, who in the vast majority hit 5ghz on kabylake

 

[humbug]

funny, I've got a 7700k at 5.1ghz, never had the heatapreader off and doesn't break past 85c after 2 hours of stress testing.....

I'm sure you're grateful to Intel for the crate of 7700K's you picked that out of, right?

 

[humbug]

Found this interesting link over on the intel sub reddit. It talks about how instructions are executed on Ryzen processor.

Seems that in theory Ryzen could have a better IPC than intel in certain workloads. Potentially something that could be changed if AMD could get there own compiler out there.(?)
Also the reason why AMD SMT is better than intels:


Recommend reading the whole thing.

Subed. i'll read in later... thx

 

[zornyan]

Found this interesting link over on the intel sub reddit. It talks about how instructions are executed on Ryzen processor.

Seems that in theory Ryzen could have a better IPC than intel in certain workloads. Potentially something that could be changed if AMD could get there own compiler out there.(?)
Also the reason why AMD SMT is better than intels:


Recommend reading the whole thing.

isn't this the same as their gpus though, look back at the fury, even polaris. they've always had more tflops and compute power than nvidia, whilst falling a decent amount behind in real world performance.

 

[zornyan]

I'm sure you're grateful to Intel for the crate of 7700K's you picked that out of, right?

if only Amazon let me walk round their warehouses eh? a man can dream

 

[=XDC=FluphyBunny]

.............

personally I'm after the 7820x anyway, 5ghz on an 8 core just seems absolutely insane to me, glad to see Intel kicking amds arse as usual

Do you get paid per word?

Will be interesting to see the reviews to see just where AMD/Intel are at.

edit: actually it will be more interesting to see the user reviews.

 

[Panos]

clock for clock ryzen is 10% slower than skylake.

Which Ryzen, with what Ram speed, and which AGESA version? Are we talking about AGESA 1006 with 3600mhz CL16 ram, or AGESA 1000 with 2333CL18 Ram?
(which was used on most of the early benchmarks)
The performance difference between the two is 15% in fps/pure performance.

The Ryzen 9 16core AMD had at Computex, was running at stock speeds, and it's performance was 2.77x that of the similarly clocked Ryzen 7.
(on performance benchmarks like handbreak, not the gaming one)

Because of the scaling, we know at least the base difference with the Ryzen 7 should be 2x due to higher core count which is double.
However that extra 77% EXTRA PERF clock for clock is there and probably is coming from somewhere else.

Most bet is due to quad channel ram, allowing Infinity fabric run faster between the CCXs, I say also because the system was strapped with Vega GPUs.
Something Raj has mentioned since January 2017 CES. That a system with Ryzen + Vega will build an ecosystem that provide more performance overall than each part running on it's own.

Also lets not forget most of the mobo manufacturers went cheap for the X370 platform, not knowing the uptake. And they couldn't manage to fill the backorders......

Closing. We are talking about HEDT market. I want core count at reasonable price, the 10% IPC different you write all the time, is meaningless when the core count is 60% higher.

And I write again. Having moved from [email protected] to 6800@4Ghz, even single thread games run better, especially when I more items running on the background like TS and browser streaming radio or playing youtube. While gaming.
Let alone multicore games, the different is much bigger.

(And as you see I do not write theories, or rumours like the Ryzen 9s overclock 4.2 easily due to more mature process, or that the X399 Asus board was spotted with the onboard speed metre above 5Ghz).

 

[Chuk_Chuk]

isn't this the same as their gpus though, look back at the fury, even polaris. they've always had more tflops and compute power than nvidia, whilst falling a decent amount behind in real world performance.

AMD GPUs where hamstrung by the hardware scheduler and DX11; The fury wasn't balanced properly so lead to poor performance. Nvidia shined in the DX11 days because they did a lot of there scheduling in software (It's also where they gained their efficiency from). Its also where the finewine stuff comes in. AMD cards are less reliant on drivers for performance, hence why they appear age better. Nvidia just stops optimising there old GPUs.

 

[Chuk_Chuk]

The Ryzen 9 16core AMD had at Computex, was running at stock speeds, and it's performance was 2.77x that of the similarly clocked Ryzen 7.
(on performance benchmarks like handbreak, not the gaming one)

It was in Blender, I don't think they showed handbrake in the event but I could be mistaken.

 

[Panos]

It was in Blender, I don't think they showed handbrake in the event but I could be mistaken.

I could be at fault, since have slept since.

However the question remains. Where that 77% extra perf came from.....

 

[zornyan]

Which Ryzen, with what Ram speed, and which AGESA version? Are we talking about AGESA 1006 with 3600mhz CL16 ram, or AGESA 1000 with 2333CL18 Ram?
(which was used on most of the early benchmarks)
The performance difference between the two is 15% in fps/pure performance.

The Ryzen 9 16core AMD had at Computex, was running at stock speeds, and it's performance was 2.77x that of the similarly clocked Ryzen 7.
(on performance benchmarks like handbreak, not the gaming one)

Because of the Infinity Fabric, we know at least the base difference with the Ryzen 7 should be 2x due to higher core count which is double.
However that extra 77% EXTRA PERF clock for clock is coming from somewhere else.

Most bet is due to quad channel ram, allowing Infinity fabric run faster between the CCXs, I say also because the system was strapped with Vega GPUs.
Something Raj has mentioned since January 2017 CES. That a system with Ryzen + Vega will build an ecosystem that provide more performance overall than each part running on it's own.

Also lets not forget most of the mobo manufacturers went cheap for the X370 platform, not knowing the uptake. And they couldn't manage to fill the backorders......

Closing. We are talking about HEDT market. I want core count at reasonable price, the 10% IPC different you write all the time, is meaningless when the core count is 60% higher.

And I write again. Having moved from [email protected] to 6800@4Ghz, even single thread games run better, especially when I more items running on the background like TS and browser streaming radio or playing youtube. While gaming.
Let alone multicore games, the different is much bigger.

every test had the ryzen using similar ram speeds to the 6900k /5960xs they were benchmarking, usually 3000mhz on all the setups, although technically in a workstation where overclocking isn't used, it'll be 2133mhz the base speed on all the systems, which if I'm not mistaken will actually hurt ryzen more than Intel due to infinity fabric no?

motherboard suppliers did mess up, but that sounds like it was partly amd's fault from what they said, part of the supply issue with some like the asus crosshair was down to asus having to pull back a lot of stock to get boards to stop bricking themselves, which in turn was because amd apparently have them 0 time with the cpus to actually get a working bios ready for launch.

gotta call ******** on the 4ghz 6800k performing better than a 4.8ghz 6700k, even ARMA (an actual single threaded game) ran at about 45fps with a my 4ghz 1800x, and averages 75-80 with my 7700k at 5.1ghz.

the 1800x had 1 core at 100% constantly and just couldn't keep up at all

haven't found a game yet that performs better on the 1800x than the 7700k

 

[Doobedoo]

haven't found a game yet that performs better on the 1800x than the 7700k

In this article, we focused entirely on gaming using the new AMD Balanced power plan. Of the 14 games we tested, many of the tests the AMD Ryzen 1700X was able to hold it’s own against the Intel Core i7-7700K. There were times where the AMD Ryzen fell behind the Intel i7 by as much as 10%. However, the inverse is also true, sometimes the AMD Ryzen 1700X took the lead over the Intel i7.

On average there was between 1FPS and 5FPS difference between the two test systems across most of the games we took a look at. The one outlier was FarCry 4, where at 1080P where the Intel i7-7700k had a commanding lead with 24FPS! That’s a huge difference, just to verify that difference, we actually tested that game a few extra times. Likewise, in this round of testing, Grand Theft Auto V received nearly even results between the two test systems, again we tested this extra times. In both cases, the results were the same no matter how many times we tested it. If you run at 2560×1440 or higher, the difference between the two systems are negligable between the two systems.

http://www.legitreviews.com/intel-core-i7-7700k-versus-amd-ryzen-1700x-14-game-cpu-showdown_192508/8
http://www.legitreviews.com/intel-core-i7-7700k-versus-amd-ryzen-1700x-14-game-cpu-showdown_192508/8
I'm not sure if everyone is having the same results that you did.

 

[zornyan]

In this article, we focused entirely on gaming using the new AMD Balanced power plan. Of the 14 games we tested, many of the tests the AMD Ryzen 1700X was able to hold it’s own against the Intel Core i7-7700K. There were times where the AMD Ryzen fell behind the Intel i7 by as much as 10%. However, the inverse is also true, sometimes the AMD Ryzen 1700X took the lead over the Intel i7.

On average there was between 1FPS and 5FPS difference between the two test systems across most of the games we took a look at. The one outlier was FarCry 4, where at 1080P where the Intel i7-7700k had a commanding lead with 24FPS! That’s a huge difference, just to verify that difference, we actually tested that game a few extra times. Likewise, in this round of testing, Grand Theft Auto V received nearly even results between the two test systems, again we tested this extra times. In both cases, the results were the same no matter how many times we tested it. If you run at 2560×1440 or higher, the difference between the two systems are negligable between the two systems.

http://www.legitreviews.com/intel-core-i7-7700k-versus-amd-ryzen-1700x-14-game-cpu-showdown_192508/8
I'm not sure if everyone is having the same results that you did.

I play a fairly diverse set of games the worst ones off the top of my head were

rainbow six siege (maps like favela dropped as low as 70fps compared to a solid 100+ on 7700k)

the division (100fps locked on 7700k in pvp, about 80 with 1800x)

ARMA 3 (45ish fps on the ryzen, 75+ on the i7)
day-z (same as above)
fallout 4 (locked to 75fps due to game engine logic, but on the 1800x falls down as low as 50fps in cpu intensive areas)

elder scrolls online, seemed basically unplayable compared to the 7700k, since it's a heavily single threaded game cyrodil/PvP just feels like a stuttery mess on the ryzen.

and that's just the games I've been playing over the last few weeks.

 

[Panos]

gotta call ******** on the 4ghz 6800k performing better than a 4.8ghz 6700k, even ARMA (an actual single threaded game) ran at about 45fps with a my 4ghz 1800x, and averages 75-80 with my 7700k at 5.1ghz.

the 1800x had 1 core at 100% constantly and just couldn't keep up at all

haven't found a game yet that performs better on the 1800x than the 7700k

Arma 3 is a terrible game to draw comparisons. The engine is crap, with 0 optimizations since forever.

 

[Curlyriff]

So lets see, just to provide clarity on what I am getting.

With an R7 at 3.7GHz and 2400Mhz RAM I am finding that it is pretty similar/same as the i7 7700K at 4.8GHz with 3200Mhz RAM.

Siege, Division & Fallout 4 are within a few fps of each other at each resolution accordingly.
No idea on Dayz as I don't play that
Arma is awful because it uses 1 core. That isn't a knock to AMD or a massive plus to Intel really since it is just a really awful software/engine limitation
Same with Elder scrolls although it still holds 40+ fps from all accounts which isn't a stuttery mess but again single core rubbish.

Sorry but you are clearly having issues for some reason that neither myself or others have as demonstrated. You can see results in video for the division & fallout 4 on YouTube also showing the same experience I have.

 

[zornyan]

So lets see, just to provide clarity on what I am getting.

With an R7 at 3.7GHz and 2400Mhz RAM I am finding that it is pretty similar/same as the i7 7700K at 4.8GHz with 3200Mhz RAM.

Siege, Division & Fallout 4 are within a few fps of each other at each resolution accordingly.
No idea on Dayz as I don't play that
Arma is awful because it uses 1 core. That isn't a knock to AMD or a massive plus to Intel really since it is just a really awful software/engine limitation
Same with Elder scrolls although it still holds 40+ fps from all accounts which isn't a stuttery mess but again single core rubbish.

Sorry but you are clearly having issues for some reason that neither myself or others have as demonstrated. You can see results in video for the division & fallout 4 on YouTube also showing the same experience I have.

fallout 4

https://youtu.be/JGuKdALBSDY

Intel averages 80fps, amd 50fps, right off the bat 30 seconds in the ryzen is at 50fps whilst the Intel is holding 75+

this is a 4ghz 1700 vs a 4ghz 6700k, so clearly you was the one having issues with your setup.

http://www.gamersnexus.net/hwreview...iew-premiere-blender-fps-benchmarks?showall=1

and this shows how the 1800x even at 4ghz barely manages to equal an i5.

what resolution you playing at with what gpu btw? I'm using a titan Xp at 2ghz (hybrid installed) and 2560x1080 or 3440x1440 (I normally downscale for higher fps)

 

[Rroff]

His sig says SLI 980 - given how poorly a lot of recent titles play with SLI that would likely go a long way to explain why he doesn't see much difference between GPUs if that is what he is using.

 

[mmj_uk]

A CPU with faster cores will usually always get higher fps in games assuming that it's not GPU bottlenecked, unless it has too few cores and is over-utilised. There's no perfect setup but the best system is a balance between the speed of the cores and how many of them there are. Going to either extreme ie. 16 cores at 2ghz or 2 cores at 5ghz is going to be detrimental to performance. It's all about balance.. I think in the next year or so the best balance will be Intel CPU's with 6-8 cores clocked at around 4.5ghz. A 16 core Threadripper at 4ghz will be no better than an 8 core Ryzen at 4ghz for gaming.

 

[faceman123]

i wonder how many generations AM4/THREADRIPPER and x299 will last. i see Intel making major adjustments to their roadmap