Ryzen-A Fail for Gamers?

[StinkyPinky]

The 6/12 CPU will be the interesting one for me.

 

[Glo.]

I have noticed the Ryzen benchmarks are not pushing it's limits at all, in fact it's not even touching the sides.
I don't know whether it's a game driver issue or what, very strange if you ask me.

I have yet seen a game that loads all of the threads in the Ryzen CPU to the limit. The game is simply not able to push enough data through the CPU, because it does not know how to do so. The biggest fault of AMD, and Ryzen for this matter is that it is... new uArchitecture. And this is the key to understanding of what is happening with the CPUs.

 

[Geegeeoh]

Any chance of an i5 Coffee Lake 6C/6T later this year?

 

[inf64]

Is that the test where they used the 2500k with 1333Mhz memory? Because if it is, people should really stop throwing it around.

Nope. They used maximum officially supported RAM for each platform. 1600Mhz Cas9 DDR3 for SB/IB which is 20% faster than 1333Mhz you imply they used.
https://www.computerbase.de/2015-10/prozessoren-benchmarks-testsystem-amd-intel-2015/

Testsysteme für AMD

Plattform Sockel CPUs Mainboard (BIOS) Speicher
High-End AM3+ Vishera ASRock 990FX Killer (P1.10)
Asus Crosshair V Formula-Z für FX-9590 (2201) 8 GByte DDR3-1866 (10-10-10-24-1T)
Mainstream FM2+ Kaveri Asus A88X-Plus 8 GByte DDR3-2133 (10-11-11-30-1T)

Testsysteme für Intel

Plattform Sockel CPUs Mainboard (BIOS) Speicher
High-End LGA 2011-3 Haswell-E Gigabyte X99-UD7 WiFi (F9a) 16 GByte DDR4-2133 (15-15-15-35-2T)
Mainstream LGA 1151 Kaby Lake Asus Z270-A Prime (0603) 8 GByte DDR4-2400 (15-15-15-35-1T)
Mainstream LGA 1151 Skylake Asus Z170-Deluxe (0504) 8 GByte DDR4-2133 (15-15-15-35-1T)
Mainstream LGA 1150 Haswell, Broadwell MSI Z97S SLI Krait Ed. (A05) 8 GByte DDR3-1600 (9-9-9-24-1T)
Mainstream LGA 1155 Sandy Bridge, Ivy Bridge Asus P8Z77-V Pro (2003) 8 GByte DDR3-1600 (9-9-9-24-1T)

Edit:

According to intel's ark website both 2500K and 2600K officially support maximum of 1333Mhz DDR3, so computerbase used overspecced DDR3, which is fine.

 

[psolord]

Just read somewhere how CPU/GPU bottleneck works then come back.If he is already on CPU bottleneck with 1070 then faster GPU wont help.Only OC cpu will help.
7700k without GPU bottleneck is able push 120-130fps in that part.
https://youtu.be/tbGT-u4i3EY?t=720
Why is here so many people whos dont know anything how CPU/GPU bottleneck works?

I am with you one this one, without that meaning that Ryzen sucks.

It's too early, too new and runs on legacy mode, since there is no other mode available. If AMD will prevail or not remains to be seen. If someone wants to help AMD with his wallet, it's his matter entirely.

CD Project Red is one of the better devs out there, so maybe they will fix quite some performance.

Also we don't know what Nvidia will be able to do with their drivers and the Ryzen's one million threads.

Also you are comparing 60fps on the 1070 with hairworks, to 120fps on a Titan with no hairworks. Not the same thing. I see what you mean and you have a point, but it's not entirely correct. Still does not paint a nice picture, especially since W3 is my favorite game.

 

[psolord]

Nope. They used maximum officially supported RAM for each platform. 1600Mhz Cas9 DDR3 for SB/IB which is 20% faster than 1333Mhz you imply they used.
https://www.computerbase.de/2015-10/prozessoren-benchmarks-testsystem-amd-intel-2015/

This link is from 2015.

Where are the specs of the new test?

 

[inf64]

This link is from 2015.

Where are the specs of the new test?

Their old platform didn't change as far as I know. Also max. supported (officially) DDR3 for SB is 1333Mhz, so they did intel a favor there.

 

[psolord]

Their old platform didn't change as far as I know. Also max. supported (officially) DDR3 for SB is 1333Mhz, so they did intel a favor there.

I read in this very forum that they used 1333Mhz memory for the test. I am not sure which is correct.

Still people seem to get angry if reviewers are not using 10Ghz memory for Ryzen, but when it comes to Intel, this is their official supported RAM speed? There is something wrong with how people are expecting AMD and Intel to be treated.

Ok I am with the underdog too, but let's not get crazy. I run my 2500k with 2133Mhz RAM just fine.

 

[Puffnstuff]

The 6/12 CPU will be the interesting one for me.

If it clocks really well then it will have the best ROI of the lot.

 

[guachi]

Wait , are you saying that Ryzen will fail to achieve higher fps in 2 years time when coupled with more powerful(nextgen) GPU ?
If that is what you are saying, you have to look back and see how that worked out for 2500K Vs 8350. Not too good for 2500K .
Ryzen will have no issues with new, more powerful GPUs. The 4C/8T chip like 7700K on the other hand. It might end up like that 2500K if we are to judge from the past experiences.

It isn't just the 2500K vs. 8350 that shows a change. If you look at the graphs you'll see the 2600K pull away from the 2500K as well. In other words, it's not an AMD vs. Intel thing. It's a more cores/threads vs. fewer cores/threads. It's just that the change isn't as dramatic with the 2600K as it was always ahead of the 2500K.

With CPU speeds advancing so slowly and people keeping their CPUs longer, the only way we'll see advancements is if software developers use more cores. The same holds true for gaming, I think. I'm not a fan of BF 1, but that game seems to be showing the way forward - good multi-threading to take advantage of all the CPU horsepower and keep FPS up.

I wouldn't be surprised to see 15-30% gaming improvement vs. the 7700K (or any 4/8 processor) just through fixes (BIOS/OS/game patches) and just more games that can take advantage of more threads. That's a speed boost the 7700K isn't going to have. Some playthroughs already show it at 90+% CPU usage.

 

[Head1985]

I am with you one this one, without that meaning that Ryzen sucks.

It's too early, too new and runs on legacy mode, since there is no other mode available. If AMD will prevail or not remains to be seen. If someone wants to help AMD with his wallet, it's his matter entirely.

CD Project Red is one of the better devs out there, so maybe they will fix quite some performance.

Also we don't know what Nvidia will be able to do with their drivers and the Ryzen's one million threads.

Also you are comparing 60fps on the 1070 with hairworks, to 120fps on a Titan with no hairworks. Not the same thing. I see what you mean and you have a point, but it's not entirely correct. Still does not paint a nice picture, especially since W3 is my favorite game.

We both know how CPU/GPU bottleneck works(i think we are only one on this forum btw) i have just tried show some cpu limit to this people here, but they are just lost case.They know pretty much zero about how this thing works.
I am not saying ryzen is super bad, but intel is just better in gaming.Also witcher uses more than 8 cores so i was expecting just like crysis3 that ryzen will be faster than 7700k but its not.Its slower in both.

btw i hope you will buy ryzen so we can make some tests just like we did it with 2500k vs 6700k

 

[french toast]

We both know how CPU/GPU bottleneck works(i think we are only one on this forum btw) i have just tried show some cpu limit to this people here, but they are just lost case.They know pretty much zero about how this thing works.
I am not saying ryzen is super bad, but intel is just better in gaming.Also witcher uses more than 8 cores so i was expecting just like crysis3 that ryzen will be faster than 7700k but its not.Its slower in both.

btw i hope you will buy ryzen so we can make some tests just like we did it with 2500k vs 6700k

Trust me everyone in this forum fully understand gpu bottlenecking, its your side that fails to grasp the implications of a brand new architecture with a completely different cache system, not to mention completely different data fabric and core layout.
All software and especially games have been heavily optimized for core architecture over the last 8 years, the bios of AM4 for ryzen is very complex as ryzen is effectively an SOC, mobo makers didnt get the final microcode untill 3 weeks until launch, hence why its broken.
Likewise windows have not ben patched yet, there is at least one bug that affects performance mode, affecting gpu to cpu communication and latency, many others including probably the scheduler, all these things make HUGE difference, way more than yourself/sweepr and likes realise (or want to acknowledge )

Continually posting dodgy un verifiable youtube comparisons only makes worse.
Intel make excellent processors, lets wait until ryzen gets optimised first before igniting the fan wars.

 

[yepp]

We both know how CPU/GPU bottleneck works(i think we are only one on this forum btw) i have just tried show some cpu limit to this people here, but they are just lost case.They know pretty much zero about how this thing works.

Yeah and lets ignore the fact the Ryzen + 1070 was running Hairworks while DigitalFoundry Titan XP had Hairworks off. Lets not even entertain the question why DigitalFoundry would disable Hairworks even on a Titan XP for a CPU performance run.

Romans 1:22 Professing themselves to be wise, they became fools.

 

[unseenmorbidity]

We both know how CPU/GPU bottleneck works(i think we are only one on this forum btw) i have just tried show some cpu limit to this people here, but they are just lost case.They know pretty much zero about how this thing works.
I am not saying ryzen is super bad, but intel is just better in gaming.Also witcher uses more than 8 cores so i was expecting just like crysis3 that ryzen will be faster than 7700k but its not.Its slower in both.

btw i hope you will buy ryzen so we can make some tests just like we did it with 2500k vs 6700k

There is some irony for you.... A bottleneck is a laymen's term that is used to more or less help people visualize and understand what is happening. In actuality it is a gross oversimplification of a very complicated choreography between hardware.

Here is an example,

Run TW3 on an i3 6100 + slow DDR4.
Now, run TW3 on an i5 6500 + slow DDR4

The i5 will handily win, therefore it is a CPU bottleneck.

Then Run TW3 on an i3 6100 + Slow DDR4.
Now, run TW3 on an i3 6100 + Fast DDR4

The fast DDR4 enables the i3-6100 to handily win. Even nearly matching the i5 6500 with slow DDR4.


So, is it a memory bottleneck or CPU bottleneck? Like I said, oversimplification that helps get the general idea across.


EDIT: Here is a video, if you don't believe me.

 

[LTC8K6]

Any chance of an i5 Coffee Lake 6C/6T later this year?

Doesn't seem like it. It was the CL mobile chips coming out this year, anyway. And even they look delayed to at least 4Q.

CL Desktop will be 2018 at the earliest.

 

[Head1985]

There is some irony for you.... A bottleneck is a laymen's term that is used to more or less help people visualize and understand what is happening. In actuality it is a gross oversimplification of a very complicated choreography between hardware.

Here is an example,

Run TW3 on an i3 6100 + slow DDR4.
Now, run TW3 on an i5 6500 + slow DDR4

The i5 will handily win, therefore it is a CPU bottleneck.

Then Run TW3 on an i3 6100 + Slow DDR4.
Now, run TW3 on an i3 6100 + Fast DDR4


The fast DDR4 enables the i3-6100 to handily win. Even nearly matching the i5 6500.

So, is it a memory bottleneck or CPU bottleneck? It's certainly not both, because the i5 doesn't need fast DDR4 to trounce the i3.

Like I said, oversimplification that helps get the general idea across.


EDIT: Here is a video, if you don't believe me.

I know that.Also fast memory which can run at 4Ghz is strong point of skylake/kabylake.Ryzen on other hand have weak IMC with high lantency.
Btw here is memory speed scaling with ryzen.Not much better in witcher3
https://www.purepc.pl/procesory/premiera_i_test_procesora_amd_ryzen_r7_1800x_dobra_zmiana?page=0,43

And here are results after oc(note that they bottleneck skylake/kaby with 2133mhz ram in that review but still its faster than ryzen)
https://www.purepc.pl/procesory/premiera_i_test_procesora_amd_ryzen_r7_1800x_dobra_zmiana?page=0,40

 

[leoneazzurro]

I know that.Also fast memory which can run at 4Ghz is strong point of skylake/kabylake.Ryzen on other hand have weak IMC with high lantency.
Btw here is memory speed scaling with ryzen.Not much better in witcher3
https://www.purepc.pl/procesory/premiera_i_test_procesora_amd_ryzen_r7_1800x_dobra_zmiana?page=0,43

And here are results after oc(note that they bottleneck skylake/kaby with 2133mhz ram in that review but still its faster than ryzen)
https://www.purepc.pl/procesory/premiera_i_test_procesora_amd_ryzen_r7_1800x_dobra_zmiana?page=0,40

They should have insane issues with their Ryzen platform as theirs is the only review I've seen so far to have such bad results in Witcher 3 (Barely faster than an FX 8350 - that result only should show you that those results must be taken with a HUGE grain of salt).
https://www.computerbase.de/2017-03/amd-ryzen-1800x-1700x-1700-test/4/#diagramm-the-witcher-3-fps
https://www.kitguru.net/components/cpu/luke-hill/amd-ryzen-7-1800x-cpu-review/9/

 

[moonbogg]

Wrong.
If GTX1070 with ryzen have 80-90% GPU load=its CPU bottleneck.Even if you buy 10x faster card FPS will be same in that scene where GPU load is 80%.In this case around 60fps.
If GTX1070 runs at 99% whole time=Its GPU bottleneck.You can buy faster card and you will have more fps like we have seen with GP102 where 7700k have in that scene 120-130fps.

I hope you will now understand how CPU/GPU bottleneck works.This is also why test in 1080p with TITANXP=it will show how cpu will work with faster card in furure.Because TITANXP will be mainstream 2years from now and with ryzen you will have same fps in that scene with GTX1070 and with much faster card like GTX2080 or whatever.

What are you doing? I completely agreed with you. I know exactly how CPU bottlenecks work and the Ryzen simply isn't good enough for me. On top of that it clocks like crap. I'm not limiting my game performance to under 100fps with a Ryzen when I could simply buy a 6800K and max out my monitor's refresh and be done with it. I'm not willing to gamble my performance on more of AMD's promises for improved performance "soon". I'm also not willing to wait until 8 cores start performing better in games than Intel 4/6 cores do right now. It will be 5 years or more before 8 cores are actually a thing in gaming. It will take for absolute ever. A quad with HT and high clocks still blows everything else away with very few exceptions. I don't want to spend money on a Ryzen so I can enjoy super high performance in Cinebench. I'm done with this mess of a launch and will probably never use AMD in my life. They blew it.

Look at this amazing review. They included a 2700K@4.6, which is pretty representative of my current chip. Ryzen is not an upgrade to Sandybridge gamers. It just isn't. The 2700k is either pretty much tied or beats an 1800x, even when 1800x is "overclocked". 6800K blows it out of the water. 7700k is faster, but 6 cores on a 6800K is future proof enough for future API's and threaded games. 6800K is $360 at microcenter right now and I'm buying one. Not waiting until August for Skylake-X either. Screw this mess.

https://www.kitguru.net/components/cpu/luke-hill/amd-ryzen-7-1800x-cpu-review/10/

 

[Head1985]

They should have insane issues with their Ryzen platform as theirs is the only review I've seen so far to have such bad results in Witcher 3 (Barely faster than an FX 8350 - that result only should show you that those results must be taken with a HUGE grain of salt).
https://www.computerbase.de/2017-03/amd-ryzen-1800x-1700x-1700-test/4/#diagramm-the-witcher-3-fps
https://www.kitguru.net/components/cpu/luke-hill/amd-ryzen-7-1800x-cpu-review/9/

Wrong...if you test in novigrad then results are similar
http://pclab.pl/art72996-15.html
http://www.pcgameshardware.de/Ryzen-7-1800X-CPU-265804/Tests/Test-Review-1222033/

 

[leoneazzurro]

Please show me where "novigrad" is written in the pcgameshardware review.
Also, that last review shows the FX8370 again on par, which by common sense cannot really be.
Please note at the end also "results are not comparable" as they use a GTX980 whereas the other reviews I linked to you use faster graphic cards. So they should be less and less GPU limited.
But, of course, it depends on what are you trying to demonstrate by cherrypicking.

 

[Glo.]

I cannot wrap my head around, how come people can look at results, and draw completely wrong conclusions about them.

Perception of the CPU drives their conclusions?

F****** FX 8370 is few % slower than Ryzen in the tests. Nothing red does not light up in your minds, guys?

It COMPLETELY defies logic. The numbers defy logic, especially when you compare with professional calculations, and scientific/HPC workloads.

Pull yourselves together, guys. You show complete incompetence in terms of understanding what is happening.

 

[Puffnstuff]

I consider any OC'd tests to be outliers and as such I ignore them.

 

[Glo.]

Let me show you something.

Project Cars: 6600K +GPU loads: 70-75% for CPU, 90-95% for GPU, 1700+GPU loads: 25-30(!)% CPU, 97-99% GPU(!)
Battlefield 1: 6600K: CPU load: 97-100%, GPU load: 70-75%, 1700: 44-50% CPU load, 65-75% GPU load(the f***?)
The Witcher 3, low settings, 6600K: 80-90% CPU load, 75-95% GPU load. Ryzen 7 1700: 35-49% CPU load, 75-99% GPU load.
Far Cry Primal: 6600K: 63-70% CPU load, 79-90% GPU load, 1700: 20(!)-30% CPU load, 75-90% GPU load.

It goes on, and on, and on...

Ill tell you exactly what is happening. The CPU is not fully utilized in games. Even GTAV which is heavy on CPU uses between 20 and 40% of its capablities.

It appears that 90% of youtube reviewers are blatantly incompetent in interpreting results, and what is actually happening here. And that is driving, again, AMD mindshare as a brand.

 

[Head1985]

Let me show you something.

Project Cars: 6600K +GPU loads: 70-75% for CPU, 90-95% for GPU, 1700+GPU loads: 25-30(!)% CPU, 97-99% GPU(!)
Battlefield 1: 6600K: CPU load: 97-100%, GPU load: 70-75%, 1700: 44-50% CPU load, 65-75% GPU load(the f***?)
The Witcher 3, low settings, 6600K: 80-90% CPU load, 75-95% GPU load. Ryzen 7 1700: 35-49% CPU load, 75-99% GPU load.
Far Cry Primal: 6600K: 63-70% CPU load, 79-90% GPU load, 1700: 20(!)-30% CPU load, 75-90% GPU load.

It goes on, and on, and on...

Ill tell you exactly what is happening. The CPU is not fully utilized in games.

From when games uses 16cores?

 

[USER8000]

What are you doing? I completely agreed with you. I know exactly how CPU bottlenecks work and the Ryzen simply isn't good enough for me. On top of that it clocks like crap. I'm not limiting my game performance to under 100fps with a Ryzen when I could simply buy a 6800K and max out my monitor's refresh and be done with it. I'm not willing to gamble my performance on more of AMD's promises for improved performance "soon". I'm also not willing to wait until 8 cores start performing better in games than Intel 4/6 cores do right now. It will be 5 years or more before 8 cores are actually a thing in gaming. It will take for absolute ever. A quad with HT and high clocks still blows everything else away with very few exceptions. I don't want to spend money on a Ryzen so I can enjoy super high performance in Cinebench. I'm done with this mess of a launch and will probably never use AMD in my life. They blew it.

Look at this amazing review. They included a 2700K@4.6, which is pretty representative of my current chip. Ryzen is not an upgrade to Sandybridge gamers. It just isn't. The 2700k is either pretty much tied or beats an 1800x, even when 1800x is "overclocked". 6800K blows it out of the water. 7700k is faster, but 6 cores on a 6800K is future proof enough for future API's and threaded games. 6800K is $360 at microcenter right now and I'm buying one. Not waiting until August for Skylake-X either. Screw this mess.

https://www.kitguru.net/components/cpu/luke-hill/amd-ryzen-7-1800x-cpu-review/10/

Sadly,when I started talking about those Fallout 4 results on another forum,which he mentioned (and seem to correlate to some other data I saw),people attacked me and reported my posts to shut me up,since I was apparently a troll since Sandy Bridge was good enough for all games. Its getting very toxic TBH,so best to not talk about any games or situations were Ryzen might do worse than Intel in, and that is from some one who on record has said they don't want to give Intel any more money,ie,they want to support AMD. It makes me wonder whether the Tech of Tomorrow guy was correct!