[Techspot] AMD Ryzen 5 1600 vs. Intel Core i7-7800X: 30 Game Battle! [Links Fixed - Updated]

[MarkPost]

Which is even higher than I've observed. The majority of 1700 chip overclocks I've observed on forums are around 3.8Ghz and lower. 4Ghz @ 1.440v and Asus Real bench stable is quite the achievement.

Edit: You do have Ryzen, right? Try the Y-Cruncher avx stress test to see where your true clocks should be.

I've not used Y-Cruncher for my stability OCs (I normally use OCCT), but I wil try it in two of my systems (1700X and 7800X).

Meanwhile, whats the average Skylake-X true clock achievable with OC stability tools such as OCCT, Prime95 or Y-Cruncher?

 

[Zucker2k]

Are there any benchmarks that compare these two with Y-Cruncher? Would be surprised if a 1600 or 1700 pulls more than 7800X or 7820X.

Not sure, but a member here (DrMXLord?) run the tests and reported his findings. At least, another Ryzen user also had to raise voltages in order to pass the tests.

 

[Zucker2k]

Well considering TH have 7820X pulling 180.5w at stock on Prime95.....

AVX-512 ports at work, most likely.

 

[Anarchist Mae]

The only people i've seen not hit over 3.8Ghz on Ryzen, ANY Ryzen, are the ones using the stock cooler.

Speaking of which, my Arctic Cooling Liquid Freezer 360 arrived! Still waiting on the AM4 bracket however...

 

[Zucker2k]

I've not used Y-Cruncher for my stability OCs (I normally use OCCT), but I wil try it in two of my systems (1700X and 7800X).

Meanwhile, whats the average Skylake-X true clock achievable with OC stability tools such as OCCT, Prime95 or Y-Cruncher?

Varies with cooling. You should be aiming for 4.6Ghz with your 7800x, with good cooling, of course.

 

[Zucker2k]

Here's the post from member DrMrLordx on Ryzen and YCruncher: https://forums.anandtech.com/thread...iscussion-thread.2508085/page-6#post-38974890

 

[MarkPost]

Varies with cooling. You should be aiming for 4.6Ghz with your 7800x, with good cooling, of course.

In my 7800X I finally installed a TR Le Grand Macho w/ a Corsair ML-140 Pro fan. In one of my two 1700X I just installed a Noctua NH-D15S w/ a Corsair ML-140 Pro fan too (the other 1700X is with just a Lepa NEOIllusion, wich is cool )

Max stable OC (using OCCT) for my 7800X is 4.5 with auto vcore in BIOS, ~1.14v). I tried 4.6 with 1.2v but failed and temps were already insane, so I cant try a higher vcore

Max stable OC (OOCT) with the first 1700X I mentioned is 4.0 ~1.43v in BIOS. But I normally use a 3.9 ~1.36v

Lets see what happens with Y-Cruncher

 

[coercitiv]

In my 7800X I finally installed a TR Le Grand Macho w/ a Corsair ML-140 Pro fan.

What temps are you getting at 4.5Ghz under stress?

 

[Zucker2k]

In my 7800X I finally installed a TR Le Grand Macho w/ a Corsair ML-140 Pro fan. In one of my two 1700X I just installed a Noctua NH-D15S w/ a Corsair ML-140 Pro fan too (the other 1700X is with just a Lepa NEOIllusion, wich is cool )

Max stable OC (using OCCT) for my 7800X is 4.5 with auto vcore in BIOS, ~1.14v). I tried 4.6 with 1.2v but failed and temps were already insane, so I cant try a higher vcore

Max stable OC (OOCT) with the first 1700X I mentioned is 4.0 ~1.43v in BIOS. But I normally use a 3.9 ~1.36v

Lets see what happens with Y-Cruncher

That's a no no. That's rather low for the 7800x. Some reviews even managed 4.8Ghz (apparently prime stable). You may need to bang heads with someone with a similar chip and board and see where that takes you. The delta between 1.14v @ 4.5Ghz and 1.20v for that extra 100mhz seems too much. Pay attention to the other settings as well. As a general rule, you don't want to leave anything on auto.

 

[MarkPost]

What temps are you getting at 4.5Ghz under stress?

Cores, 81º the coolest, 87º the hottest. CPU package 92º max

The case is a BeQuiet Dark Base 900 Pro w/ 3 Corsair ML-140 Pro (2 up, 1 back).

All fans running ~1600rpm under stress (OCCT). Room temp is really high these days though, ~29º

 

[MarkPost]

That's a no no. That's rather low for the 7800x. Some reviews even managed 4.8Ghz (apparently prime stable). You may need to bang heads with someone with a similar chip and board and see where that takes you. The delta between 1.14v @ 4.5Ghz and 1.20v for that extra 100mhz seems too much. Pay attention to the other settings as well. As a general rule, you don't want to leave anything on auto.

I tested vcore manually too at 4.5. 1.14v in BIOS is just about the same that auto. Mesh is OCd to 2800. But 4.6 at 1.2v is a no go (tried mesh at 2800 and default). I think its a matter of high temps with these babes.

I would like to see Skylake-X users reporting their stable OCs with OCCT, Prime95, Y-Cruncher... but basically nobody report them, the more I've seen is someone with 10-15min of Prime, its really weird

 

[MarkPost]

well, this Y-Cruncher is THE HELL

With my OCCT stable OC settings (listed above, 1700X 4.0; 7800X 4.5), both rebooted running 3º test, with 1700X reaching 85º (tdie), and 7800X surpassing 100º (some cores and package)

 

[Zucker2k]

well, this Y-Cruncher is THE HELL

With my OCCT stable OC settings (listed above, 1700X 4.0; 7800X 4.5), both rebooted running 3º test, with 1700X reaching 85º (tdie), and 7800X surpassing 100º (some cores and package)

Yeah. Keep in mind that with 7800x your avx2 x2 (512) ports? may have been activated, which would suck a great amount of power at those clocks, resulting in lots of heat.

 

[MarkPost]

Yeah. Keep in mind that with 7800x your avx2 x2 (512) ports? may have been activated, which would suck a great amount of power at those clocks, resulting in lots of heat.

yes, and its a problem when I do some video encoding work. With x264 codec which uses some avx instructions, cores reach ~82º and cpu package ~88º. With x265 its worst, ~86º cores and ~92º package. Fortunately x264 is my favourite video codec, much faster than x265 and I would say with better final quality. I'm not confortable with x264 temps either, but I can live with it. At least, when summer finish temps will be better xD

 

[Anarchist Mae]

Some good news, Hardware Unboxed has purchased a new 7800X to do a retest with:

https://www.youtube.com/watch?v=gKrrCtmsmos

Of course even if it scores better it still costs twice as much at a minimum.

 

[inf64]

Some good news, Hardware Unboxed has purchased a new 7800X to do a retest with:

https://www.youtube.com/watch?v=gKrrCtmsmos

Of course even if it scores better it still costs twice as much at a minimum.

It will hardly score any better, their original results are perfectly in line with say hardware.fr's SKL-X gaming results. There is no magic pixie dust to improve the horrendous perf./$ of SKL-X in gaming.

 

[Crumpet]

Some good news, Hardware Unboxed has purchased a new 7800X to do a retest with:

https://www.youtube.com/watch?v=gKrrCtmsmos

Of course even if it scores better it still costs twice as much at a minimum.

Lets see if this one decides to commit harakiri too.

 

[Christopher Bohling]

I will admit to being stupid, but surely the games where the 7800X fares poorly and the 7700K shoots way ahead of both the 1600 and 7800X is due to the core-to-core latency deficit, right? I mean, look at the FCP and TW:W benchmarks. Techtubers keep saying that "Far Cry Primal is limited by single-threaded performance" but obviously that's not the only limiting factor, otherwise the 7800X at 4.7Ghz would be only a few % slower than the 7700K at 4.9Ghz (since we all know Skylake and Kaby Lake have a virtually identical IPC).

So the factor that seems to be making difference is the low-latency ring bus on the 7700K vs. the higher-latency mesh/fabric designs on SKL-X and Ryzen? Does that make sense?

EDIT: this would also of course be why Broadwell-E is better at gaming than Skylake-X

 

[PhonakV30]

I tested vcore manually too at 4.5. 1.4v in BIOS is just about the same that auto. Mesh is OCd to 2800. But 4.6 at 1.2v is a no go (tried mesh at 2800 and default). I think its a matter of high temps with these babes.

I would like to see Skylake-X users reporting their stable OCs with OCCT, Prime95, Y-Cruncher... but basically nobody report them, the more I've seen is someone with 10-15min of Prime, its really weird

Be carefull with your OC 7800x , here

https://www.instagram.com/p/BWysDUalr2k/

check your Pins ASAP! his voltage was 1.25v but yours at 1.45v

 

[TheELF]

I will admit to being stupid, but surely the games where the 7800X fares poorly and the 7700K shoots way ahead of both the 1600 and 7800X is due to the core-to-core latency deficit, right? I mean, look at the FCP and TW:W benchmarks. Techtubers keep saying that "Far Cry Primal is limited by single-threaded performance" but obviously that's not the only limiting factor, otherwise the 7800X at 4.7Ghz would be only a few % slower than the 7700K at 4.9Ghz (since we all know Skylake and Kaby Lake have a virtually identical IPC).

So the factor that seems to be making difference is the low-latency ring bus on the 7700K vs. the higher-latency mesh/fabric designs on SKL-X and Ryzen? Does that make sense?

EDIT: this would also of course be why Broadwell-E is better at gaming than Skylake-X

One could think so but less threads equals less communication between the few threads which makes the ring bus latency less important.

 

[Christopher Bohling]

One could think so but less threads equals less communication between the few threads which makes the ring bus latency less important.

Right, but let's say a program is designed to scale to 4 threads, but utilizes all 4 of those threads equally and frequently shares information between them. Wouldn't 4 threads with 30ns of latency see a notable performance improvement over 4 threads with 130ns of latency? (again this is a genuine question)

If that's not the reason, what other architectural difference accounts for SKL-X performing more like Ryzen and less like Kaby Lake in Far Cry Primal?

 

[MarkPost]

Be carefull with your OC 7800x , here

https://www.instagram.com/p/BWysDUalr2k/

check your Pins ASAP! his voltage was 1.25v but yours at 1.45v

errr sorry there is a typo in my previous post, vcore is 1.14 not 1.4
I'll edit it now