Ryzen-A Fail for Gamers?

[R0H1T]

Wrong. i7-5820k with just 3.3-3.6 Ghz and only 6 cores is faster in gaming at Computerbase. Many tests are GPU bound. Gaming IPC is below Haswell.

No such thing as gaming IPC, since game/OS patches, drivers, GPU, RAM & even storage make a huge difference to your gaming experience.
If we're to tie IPC loosely to each game, even with the same engine, we'll see so much variance, depending on whether it was AMD or Nvidia optimized, that the term would be nothing short a joke literally.

 

[Rngwn]

I did. Ryzen is clearly below Haswell IPC in games. It looks better in applications but it's still very inconsistent in performance and power comsumption can be really high, sometimes 50-100W above i7-7700k with the same TDP.

And Ryzen isn't AVX future proof.

http://www.tomshardware.com/reviews/amd-ryzen-7-1800x-cpu,4951-4.html

Bear in mind that the 6900k is tested with Quad-channel RAM and the Ryzen itself has crappy latency at its current form. IIRC, the DFT/FFT is memory intensive, so taking away the advantage of quad-channel and the result could be not as disastrous as this. Correct me if I'm wrong here.

 

[unseenmorbidity]

I did. Ryzen is clearly below Haswell IPC in games. It looks better in applications but it's still very inconsistent in performance and power comsumption can be really high, sometimes 50-100W above i7-7700k with the same TDP.

And Ryzen isn't AVX future proof.

Why is sweeper hiding and liking anything that seems critical of AMD?

First off, optimizations are coming, and should make a large difference. Secondly, who measures IPC from gaming!? Nonsense...

 

[ultima_trev]

Wrong. i7-5820k with just 3.3-3.6 Ghz and only 6 cores is faster in gaming at Computerbase. Many tests are GPU bound. Gaming IPC is below Haswell.

Wrong.

You clearly didn't read my post or the article: https://www.computerbase.de/2017-03/amd-ryzen-1800x-1700x-1700-test/4/

Over 80% of the 6950X in gaming performance over all, even at 720P. Please stop spreading misinformation.

 

[AMDisTheBEST]

I did. Ryzen is clearly below Haswell IPC in games. It looks better in applications but it's still very inconsistent in performance and power comsumption can be really high, sometimes 50-100W above i7-7700k with the same TDP.

And Ryzen isn't AVX future proof.

I need to correct you on tdp. 1700 has 65 watts. It is 8 cores. Each core requires power. 7700k has just 4 cores. It's tdp should be much lower. The fact that even an 8 core 1700 best it in power consumption shows ryzen is much more energy efficient.

 

[sirmo]

I need to correct you on tdp. 1700 has 65 watts. It is 8 cores. Each core requires power. 7700k has just 4 cores. It's tdp should be much lower. The fact that even an 8 core 1700 best it in power consumption shows ryzen is much more energy efficient.

Ryzen's energy power draw quckly ramps up when you hit those high clocks, but when it comes to mainstream clocks for laptop and datacenter users.. in other words when Ryzen is running in it's sweet spot clocks like 1700 is, it absolutely has more perf/watt than Intel.

 

[IEC]

TechDeals just did a look at minimum FPS:

https://youtu.be/1Ovr9RostHk?t=1564

He's supposed to do a multiplayer BF1 comparison soon. I suspect we'll see the i7-7700K choke on 64 player maps like my 6700K does on Amiens...

 

[psolord]

Checked computerbase.de and it seems their test suite of 13 games uses more games that benefit from lots of cores.

The overall ranking of the processors tested from their 1920x1080 scores (giving a 2500K an overall score of 100):
189.7 6900K
186.9 6950X
185.1 6850K
172.8 7700K
169.3 1800X
148.5 7600X
139.5 4770K
120.4 FX-9590
116.8 2600K
112.9 FX-8370
100.0 2500K

In this test suite, the 7700K and the 1800X were almost equal with the 1800X only about 2% slower overall. The $590 6950K does quite well, but I think the $500 1800X beats it handily in CPU tests.

The 1700X, while not tested by computerbase.de, should perform only a few percent worse than the 1800X for $100 cheaper. Therefore, I think the 7700K is not a good buy using these games as an example. We'll likely only see more games take advantage of >4 cores and thus give us more games where AMD outright beats the 7700K.

However, the 7600K at $240 is a great value as you get the same FPS/$ of the two FX chips but much higher FPS/$ than any of the more expensive chips.

Conclusion: Using this gaming test suite, AMD has the best gaming CPU on the market in the 1800X (and 1700X) at stock speeds. Anything else is either too expensive (6900K, 6950X, 6850K) or not so much faster that I'd risk getting a four core processor (1700K). Unless value is important, and then the 7600K is a good choice.

Are you talking about this test my friend?
https://www.computerbase.de/thema/prozessor/rangliste/#diagramm-gesamtrating-spiele-full-hd

Because if you do, I think you have confused the game only results with the overall results.

 

[Head1985]

We tested r7 1700 in another forum in games in cpu limited situations and looks like ryzen have around SB/ivy IPC.
In fallout4 6700k 4.5Ghz was 64% faster than r71700 at 4Ghz in very cpu limited situation.
82fps for 6700k and 50fps on ryzen
In crysis3 in grass level 6700k was 26% faster.105fps vs 83fps.(2500k at 4.5ghz have there 52fps)

Zen is very weak in games where is needed memory badwidth and low latency.But strong in games that dont need it like old UT99.There its faster than haswell.
Faster memory helps, but zen just have slow IMC with high latency and that killing gaming performance.

 

[unseenmorbidity]

We tested r7 1700 in another forum in games in cpu limited situations and looks like ryzen have around SB/ivy IPC.
In fallout4 6700k 4.5Ghz was 64% faster than r71700 at 4Ghz in very cpu limited situation.
82fps for 6700k and 50fps on ryzen
In crysis3 in grass level 6700k was 26% faster.105fps vs 83fps.(2500k at 4.5ghz have there 52fps)

Zen is very weak in games where is needed memory badwidth and low latency.But strong in games that dont need it like old UT99.There its faster than haswell.
Faster memory helps, but zen just have slow IPC with high latency and that killing gaming performance.

bs

Who would even test fallout4? That game runs like hot garbage...

There is no such thing as gaming ipc either...

 

[guachi]

Are you talking about this test my friend?
https://www.computerbase.de/thema/prozessor/rangliste/#diagramm-gesamtrating-spiele-full-hd

Because if you do, I think you have confused the game only results with the overall results.

No.

I typed all the game results into Excel by hand. I normalized each game to give a score of 100.0 either to the slowest CPU, which turned out to be the 2500K, for every game or to give 100.0 points to the 7700K for every game. Total points up. Divide by number of games.

This way, you could see how much faster or slower any given tested CPU was than the 2500K or the 7700K. The 1800X averaged a score of 98.3 vs. the 7700K score of 100.0. Therefore, the 1800X was 1.7% slower overall than the 7700K. Or, it was 69.3% faster than the 2500K.

You are welcome to do the math yourself if you want.

 

[dark zero]

I did. Ryzen is clearly below Haswell IPC in games. It looks better in applications but it's still very inconsistent in performance and power comsumption can be really high, sometimes 50-100W above i7-7700k with the same TDP.

And Ryzen isn't AVX future proof.

*looks at FX*

Remember that is Dual vs Quad... look at the i7 7700K.... is supposed to be WAY better than Broadwell... and look at it... Duals are supposed to NOT to be FutureProff...

Also the HEDT are rejected Xeon and they are optimized for OC...

 

[Head1985]

bs

Who would even test fallout4? That game runs like hot garbage...

There is no such thing as gaming ipc either...

fallout4 is great for testing CPU.Btw it runs pretty good if you have skylake/kaby with 3000+ memory.
I think ryzen is decent cpu for gaming.IPC like SB/ivy in worst case is not that bad.SLOW IMC with high latency killing gaming performance.

 

[alexruiz]

We tested r7 1700 in another forum in games in cpu limited situations and looks like ryzen have around SB/ivy IPC.
In fallout4 6700k 4.5Ghz was 64% faster than r71700 at 4Ghz in very cpu limited situation.
82fps for 6700k and 50fps on ryzen
In crysis3 in grass level 6700k was 26% faster.105fps vs 83fps.(2500k at 4.5ghz have there 52fps)

Zen is very weak in games where is needed memory badwidth and low latency.But strong in games that dont need it like old UT99.There its faster than haswell.
Faster memory helps, but zen just have slow IMC with high latency and that killing gaming performance.

Ok, so if "you" really tested a R7 1700, how about you share all the settings and configurations?
Meaning, board and firmware version, RAM and timings used, storage, operating system version and updates, video card and drivers, etc
Anyone can post and claim anything.

 

[unseenmorbidity]

fallout4 is great for testing CPU.Btw it runs pretty good if you have skylake/kaby with 3000+ memory.
I think ryzen is decent cpu for gaming.IPC like SB/ivy in worst case is not that bad.SLOW IMC with high latency killing gaming performance.

Anything made by bethesda runs like garbage.

For intel anyways, memory speed is basically irrelevant unless your CPU is becoming a bottleneck.

That is unless the game runs like hot garbage, like for instance Fallout4, which actually does run better with faster memory regardless if a bottleneck is there. These games are extreme outliers though.

Ryzen is unstable at best right now, particularly with memory. Obviously it isn't going to do well with Fallout4...

 

[Head1985]

Ok, so if "you" really tested a R7 1700, how about you share all the settings and configurations?
Meaning, board and firmware version, RAM and timings used, storage, operating system version and updates, video card and drivers, etc
Anyone can post and claim anything.

r7 1700 4ghz, 2933Mhz CL18-17-17-38 2T, ASUS PRIME B350-PLUS win7
see yourself
http://pctforum.tyden.cz/viewtopic.php?p=9245791#p9245791

I have 6700k 4.5Ghz, 3000mhz ram cl14-15-15-35 1t

 

[unseenmorbidity]

r7 1700 4ghz, 2933Mhz CL18-17-17-38 2T, ASUS PRIME B350-PLUS win7
see yourself
http://pctforum.tyden.cz/viewtopic.php?p=9245791#p9245791

That sure proves it...

 

[Head1985]

For intel anyways, memory speed is basically irrelevant unless your CPU is becoming a bottleneck.

Of course its cpu bottleneck scene.Why even test cpu if its not in cpu bottleneck scene?

Same in crysis3 this is test scene:here ryzen have 83fps and 2500k at 4.5ghz 52fps.6700k at 4.5ghz around 105-108fps.

 

[Head1985]

i found video with ryzen in same test scene.At stock he have around 60fps and after oc to 4Ghz around 80fps(before he start running)

 

[unseenmorbidity]

There seems to be this weird disconnect from reality from reviewers. They seem to believe their test results are somehow absolute. I feel as if they don't fully understand the environment.

Let me explain.

Gamers Nexus for example, had terrible results for Ryzen in Watch Dogs 2 at 1080p.

Now, go to Guru3D, 1080p:

http://www.guru3d.com/articles_pages/amd_ryzen_7_1800x_processor_review,20.html

See what? Completely different to Gamers Nexus.

The same for BF1.

Now you go to another review site, HardwareCanucks, and guess what, they also find excellent performance.

http://www.hardwarecanucks.com/foru...-amd-ryzen-7-1800x-performance-review-15.html

You go to Computerbase.de, and look at that..

https://www.computerbase.de/2017-03.../#abschnitt_benchmarks_mit_fps_und_frametimes

Ryzen 1800X is only 4% behind 7700K gaming at 1080p.

At 720p, it is only 9% behind. No where near the 20-30% behind scenario at your preferred reviewer.

And guess what? BF1 multiplayer, 1080p Titan XP, Ryzen is FASTER than the 7700K.

https://www.computerbase.de/2017-03.../#diagramm-battlefield-1-dx11-multiplayer-fps

FASTER.

 

[USER8000]

bs

Who would even test fallout4? That game runs like hot garbage...

There is no such thing as gaming ipc either...

Well I play Fallout 4 and games like Skyrim,so just because Ryzen does relatively poorly in those games we should ignore it?? You might want to look at how many people still play games like Fallout,Skyrim,etc - they have some of the most active modding communities of any set of games. There are other games like World of Tanks and World of Warcraft which are also very popular but don't scale well to more cores.

These are the type of games which people will spend 100s of hours playing unlike some short FPS game.

Using your logic then we should ignore any game which uses 8 cores since Intel might lose,and AMD wins??

I do want to get a Ryzen CPU this year,but already have an Ivy Bridge Core i7 so would want it to be at least some degree of upgrade over my ancient CPU,even if I am trading single threaded performance for more cores.

Making excuses for poor performance in these games won't help sell more AMD CPUs,since plenty of people will look at the benchmarks and just go and get an Intel CPU instead,and AMD needs to find a way to improve performance in them.

Just trying to silence people on forums who mention it won't mean anything since AMD does need to try and do something about it.

 

[Head1985]

There seems to be this weird disconnect from reality from reviewers. They seem to believe their test results are somehow absolute. I feel as if they don't fully understand the environment.

Let me explain.

Gamers Nexus for example, had terrible results for Ryzen in Watch Dogs 2 at 1080p.

Now, go to Guru3D, 1080p:

http://www.guru3d.com/articles_pages/amd_ryzen_7_1800x_processor_review,20.html

See what? Completely different to Gamers Nexus.

The same for BF1.

Now you go to another review site, HardwareCanucks, and guess what, they also find excellent performance.

http://www.hardwarecanucks.com/foru...-amd-ryzen-7-1800x-performance-review-15.html

You go to Computerbase.de, and look at that..

https://www.computerbase.de/2017-03.../#abschnitt_benchmarks_mit_fps_und_frametimes

Ryzen 1800X is only 4% behind 7700K gaming at 1080p.

At 720p, it is only 9% behind. No where near the 20-30% behind scenario at your preferred reviewer.

And guess what? BF1 multiplayer, 1080p Titan XP, Ryzen is FASTER than the 7700K.

https://www.computerbase.de/2017-03.../#diagramm-battlefield-1-dx11-multiplayer-fps

FASTER.

It depends on:
How fast ram they used:skylake/kaby scale extremely well with fast 3000+ ram
What GPU they used: you need TITANXP in 1080p to not be gpu bottleneck
Where you test:You can test crysis3 in test scene in grass area where we tested or you can test somewhere where even 2500k can deliver 80fps(but in grass area it will be at 30-40fps at stock)

gamernexus and techspot used both TITANXP with 3000+ fast DDR4 for skylake/kaby.
http://www.gamersnexus.net/hwreview...review-premiere-blender-fps-benchmarks/page-3
http://www.techspot.com/review/1345-amd-ryzen-7-1800x-1700x/page2.html
in both skylake/kaby is way faster.

 

[unseenmorbidity]

Well I play Fallout 4 and games like Skyrim,so just because Ryzen does relatively poorly in those games we should ignore it?? You might want to look at how many people still play games like Fallout,Skyrim,etc - they have some of the most active modding communities of any set of games. There are other games like World of Tanks and World of Warcraft which are also very popular but don't scale well to more cores.

I do want to get a Ryzen CPU this year,but already have an Ivy Bridge Core i7 so would want it to be at least some degree of upgrade over my ancient CPU,even if I am trading single threaded performance for more cores.

Using your logic then we should ignore any game which uses 8 cores since Intel might lose,and AMD wins??

The issue isn't benchmarking Fallout4. It's trying to make some nonsensical conclusion from doing so. If people are reviewing these things, I would hope they understand the shortcomings of the game and software and how that might affect the outcome.

 

[unseenmorbidity]

It depend on:
How fast ram they used:skylake/kaby scale extremely well with fast 3000+ ram
What GPU they used: you need TITANXP in 1080p to not be gpu bottleneck
Where you test:You can test crysis3 in test scene in grass area where we tested or you can test somewhere where even 2500k can deliver 80fps(but in grass area it will be at 30-40fps at stock)

gamernexus and techspot used both TITANXP with 3000+ fast DDR4 for skylake/kaby.
http://www.gamersnexus.net/hwreview...review-premiere-blender-fps-benchmarks/page-3
http://www.techspot.com/review/1345-amd-ryzen-7-1800x-1700x/page2.html
in both skylake/kaby is way faster.

No, it has to do with system stability. Lots of people tested with the 1080 and got wildly different results.

HWK had a titanX.


Not sure what computerbase.de tested with. I know they have a titanX they were using though.

 

[USER8000]

The issue isn't benchmarking Fallout4. It's trying to make some nonsensical conclusion from doing so. If people are reviewing these things, I would hope they understand the shortcomings of the game and software and how that might affect the outcome.

Its not the point though - do you honestly think I like the fact that even with my ancient CPU,lots of games don't even tax half the threads?? I know how Fallout 4 is,and how Skyrim was even using X87 instructions when it was launched,etc. I looked at the Sweclockers R7 1800X review,and my old Core i7 is faster in it in Fallout 4,and an overpriced Core i3 7350K is faster than my CPU. The Core i7 7700K is like a third faster than a R7 1800X in the game. That is in Diamond City - once you get to settlement building,let alone modding,it can really push your CPU massively.

The thing is there are plenty of games where the developers just use older engines due to cost and time reasons,and people do play them and if Intel wins massively in them,they will just buy the Intel CPU instead.

It pisses me off massively that so many largish games studios still use crappy engines,when you have people like DICE who have been pushing multi-threading since the BF2 days.