How is Ryzen?

[magomago]

Need to do some research in the next few weeks to help out a friend, but I figured prepping myself when I look at it. Is Ryzen still far behind Intel in terms of CPU benchmarks for gaming? Is it a solid choice at mid to high end gaming? Is it going to wildly vary depending on the game?

Not being a troll here, just looking for a quick pulse on things. Basically need to know if "go i5/i7 don't think twice" or "no, do your research, even at higher price points it makes a difference".

 

[hackdrag0n]

I have an 1800x, x34a and a gtx1080ti. I have exactly zero complaints about gaming performance. Not a single title I play gives me less than 60hz with everything turned up.

Would I get a few more frames with a 7700k? Probably.
Would I notice the difference if I'm not staring at an fps meter? Absolutely not.

 

[IEC]

If gaming at 1080p, the i7-7700K will beat Ryzen in max and average FPS, though the difference is typically only a few %, and almost always under 9%. At 1440p and up the differences become much smaller. At 4K, there is basically no difference.

i7 still has its place in competitive gaming, but if budget precludes a >$300 processor, the Ryzen 5 1600 or 1600X is a far better value than any i5, simply due to being 6 core/12 thread vs 4 core/4 thread.

Newer games tend to utilize more cores = more cores is becoming better for gaming, as seen in the February 2017 roundup by Computerbase.de:
https://www.computerbase.de/2017-02...-test/#abschnitt_performancerating_in_full_hd

It's a safe bet that the Ryzen 5 processors will age better than the i5s with 4 cores.

 

[scannall]

The 1600 and motherboard will perform within a few percent overall of a 7700K, and cost $249 for both. The 7700K is about $100 more just for the CPU, then add a motherboard.

 

[aigomorla]

Is it a solid choice at mid to high end gaming?

Current: Yes.

In 2 Months: My guess is leaning to No... CoffeeLake / Skylake-X is supposed to bring some great improvements which Ryzen will be hard pressed to beat in high end gaming.

Basically need to know if "go i5/i7 don't think twice" or "no, do your research, even at higher price points it makes a difference".

if you can afford to wait, wait til august and see what coffee lake brings to the board.
I saw tech sheets of a hexcore with 4.5ghz which is absolutely ground stomping insane.

This is on CoffeeLake, and if you need processing prowess, skylake-x is suposed to bring a 12core/24thread monster out.

AMD will also release threadripper, which is a 16core/32thread and a new board the x399 so in general if you can afford to wait, its best to see what heavy hitters come out in august.

 

[guachi]

The 7700k is the only gaming CPU Intel sells that I would consider if you are making your own PC.

If you have any lower budget or have any other uses for your CPU other then gaming, go Ryzen.

A 1700 is cheaper (considering motherboard and cooler) and far superior in non-gaming tasks and not far (10%) behind in gaming.

Otherwise a 1600 is an insane value.

 

[bononos]

....
Not being a troll here, just looking for a quick pulse on things. Basically need to know if "go i5/i7 don't think twice" or "no, do your research, even at higher price points it makes a difference".

Unless you have very specific needs, Ryzen is the current i5-2500k of its time. Its almost a no brainer, instead of an i5, get a 1600/x or instead of an i7 get a 1700-1800x.

 

[frozentundra123456]

For gaming, the 1600x is a very good choice for around 250.00. Above 300.00 I would go with the 7700k, or wait to see what coffee lake/Skylake X brings. Here is a summary of application and gaming performance. Unfortunately they do not show the 1600X, but due to its relatively high clockspeed, in games it is close to the 1700X/1800X, and faster at stock than the 1700. hardware.fr test . Of course, it depends on the games tested, but the 7700k is close to 20% faster than the fastest Ryzen in this suite of games.

 

[Reinvented]

I've had both a 7600k system, and now a Ryzen 1600X.

My thoughts: Depends on budget, and other things like monitor. I settled on Ryzen, because minimum frames are important too. And I wanted to future proof. Ryzen is still maturing, but don't get me wrong...Intel is still very strong. You want high end right off the bat? Go Intel. But, you pay a premium for it. Go with Ryzen if you want better minimum's, future proofing, and at a very reasonable price point.

 

[itsmydamnation]

Heres a test done by @The Stilt in a CPU limited part of hitman:

Now remember most initial gaming benchmarks where done around 24000mhz with poor latency settings, a nice 20/40% increase in min/max going to 3200mhz.

 

[TheELF]

-snip-
Now remember most initial gaming benchmarks where done around 24000mhz with poor latency settings, a nice 20/40% increase in min/max going to 3200mhz.

Yeah but also nobody is going to game at 1080p with minimum quality settings if he has a Gtx1080.
Nice boost but pretty irrelevant if it doesn't happen with better settings/lower cards as well.

(Intel guys,bust out the single threaded 480p benches now)

 

[Valantar]

Jeez, what on earth is the point of benchmarking 1080p minimum on a GTX 1080? That's bonkers, even if it's to illustrate potential differences. If your goal is to find the best grocery shopping hatchback car, comparing stripped-down drag race versions isn't going to tell you jack.

CoffeeLake / Skylake-X is supposed to bring some great improvements which Ryzen will be hard pressed to beat in high end gaming.

if you can afford to wait, wait til august and see what coffee lake brings to the board.
I saw tech sheets of a hexcore with 4.5ghz which is absolutely ground stomping insane.

I don't doubt that that will perform amazingly (just imagine a 6-core 7700K ...), but I seriously hope Intel has improved their TIM by then (or just solder the damn thing). Considering temperatures in the 80s are normal for non-delidded 7700Ks even with water cooling, I would be terrified of running a 6-core chip of a similar nature. Coffee Lake is only a process refinement, if that, and considering Cannonlake is the mobile (low-power) push, I'd guess power savings will be minimal.

 

[Malogeek]

The point of the benchmark is to show that memory speed can make a difference. It won't make such a large difference overall likely and it depends on your resolution, settings and the game itself. The same goes for Intel platform, in many situations higher speeds make little difference but has the outliers where it does.

Isn't this why we do PC optimizations and overclocking? To obtain the best performance for our hardware possible so in certain situations during a game it will make a difference, even if it's just keeping it above 60hz instead of dropping down to 55.

 

[Valantar]

The point of the benchmark is to show that memory speed can make a difference. It won't make such a large difference overall likely and it depends on your resolution, settings and the game itself. The same goes for Intel platform, in many situations higher speeds make little difference but has the outliers where it does.

Isn't this why we do PC optimizations and overclocking? To obtain the best performance for our hardware possible so in certain situations during a game it will make a difference, even if it's just keeping it above 60hz instead of dropping down to 55.

Sure, but "best case scenario" benchmarks like that without any "average/more realistic scenario" benchmarks to complement and nuance the results are nigh on worthless. That's a degree of selectivity and cherry picking that inevitably leads to misinformation spreading, regardless of the intent. Some people looking at that graph will inevitably fail to look at it thoroughly, and thus go about spouting that "Ryzen gaming performance improves by 30% with faster memory!", which is simply not true.

 

[Atari2600]

Wait a sec Valantar - that is essentially exactly what the Intel fanbois (and the idiot/shill journalists) were claiming when a 7700k beat an 1800X.

Intel is still 20% faster for gaming than AMD blah blah. (What, what you mean 1080 at min settings isn't representative...? its future proofing etc etc etc *hides brown envelope*)

 

[Dygaza]

Yeah but also nobody is going to game at 1080p with minimum quality settings if he has a Gtx1080.
Nice boost but pretty irrelevant if it doesn't happen with better settings/lower cards as well.

(Intel guys,bust out the single threaded 480p benches now)

Yeah but keep in mind that the scaling is the same in scenes where Ryzen was single thread limited , and couldn't get 60 fps. We're pretty much seeing 25% scaling from 2400MHz launch timing towards 3200 LL. So in scene where you had 50 fps before, you would get over 60 now. Ofc all games don't scale that well with memory latency, but most do (as pcper showed).

Also remember that 2400MHz memories also suffered in speed when those subtimings couldn't be adjusted. Looking at how good scaling low latency gives, I would bet that Running 2400MHz memory would with tight timings be close to equal with 3200MHz with subpar timings in this test.

Overall this AGESA update is very good, and I would really like to see most sites redoing their tests, even with lower memory speeds but proper timings.

 

[Dygaza]

Sure, but "best case scenario" benchmarks like that without any "average/more realistic scenario" benchmarks to complement and nuance the results are nigh on worthless. That's a degree of selectivity and cherry picking that inevitably leads to misinformation spreading, regardless of the intent. Some people looking at that graph will inevitably fail to look at it thoroughly, and thus go about spouting that "Ryzen gaming performance improves by 30% with faster memory!", which is simply not true.

The Stilt's point with the graph was to show balance between bandwidth versus latency. You can clearly see Latency makes more difference than bandwidth. These LL data sets have also subtimings changed that weren't possible before latest AGESA updates AFAIK.

 

[itsmydamnation]

Yeah but keep in mind that the scaling is the same in scenes where Ryzen was single thread limited , and couldn't get 60 fps. We're pretty much seeing 25% scaling from 2400MHz launch timing towards 3200 LL. So in scene where you had 50 fps before, you would get over 60 now. Ofc all games don't scale that well with memory latency, but most do (as pcper showed).

Also remember that 2400MHz memories also suffered in speed when those subtimings couldn't be adjusted. Looking at how good scaling low latency gives, I would bet that Running 2400MHz memory would with tight timings be close to equal with 3200MHz with subpar timings in this test.

Overall this AGESA update is very good, and I would really like to see most sites redoing their tests, even with lower memory speeds but proper timings.

Normally i would 100% agree but we haven't really seen any great latency vs throughput tests on Zen given the fabric is tied to mem clock.

 

[Dygaza]

Normally i would 100% agree but we haven't really seen any great latency vs throughput tests on Zen given the fabric is tied to mem clock.

True, these few individual tests are everything we got so far. Well hopefully when new BIOS versions with new AGESA comes out officially, we'll see some sites actually retests Ryzen memory.

Here is another test , this time from Witcher 3 by Keketin.

https://bbs.io-tech.fi/attachments/witcher-3-memtestv3-png.21045/

Original post (in finnish)
https://bbs.io-tech.fi/threads/amd-...ukset-ja-kokemukset.14849/page-52#post-997971

Edit. Pcper actually had good article where they tested mem latency effect on different games.

https://www.pcper.com/reviews/Proce...act-Weak-1080p-Gaming/Timings-timings-timings

 

[parvadomus]

Saw this review the other day: https://www.eteknix.com/fierce-pc-imperial-hive-review/

Personally, I see the 7770K a bad choice since it has zero upgrade path. AM4 should support Zen+ cpus, and even the ryzen 1700 make it look bad at anything other than gaming, where the difference is unnoticeable.

 

[Topweasel]

Saw this review the other day: https://www.eteknix.com/fierce-pc-imperial-hive-review/

Personally, I see the 7770K a bad choice since it has zero upgrade path. AM4 should support Zen+ cpus, and even the ryzen 1700 make it look bad at anything other than gaming, where the difference is unnoticeable.

What this guy says.

The 7700k Is really really really good at games. At least most of them. It's combination of increased IPC and epic clockspeeds (even at stock). Will make most 2016 games and older run better than you will ever see them run in the future. But that is where it stops. We are already seeing an increase in thread usage with DX12 and Vulcan where the developers don't specifically write for each thread. As usage goes up the performance of the 7700 goes down. Ryzen at 6 or 8 cores already does better at high rez and is within 10% on low rez. Anywhere else Ryzen is hands down better to the tune of nearly twice as fast in work that is well threaded. This is the last 4c8t i7 desktop CPU, 90% sure its the last CPU for this platform, and just both in resources and ideology it is EOL.

Ryzen gives you twice the resources and a platform that should get at least two and probably 3 refreshes.

 

[dark zero]

Also Ryzen can be used OUTSIDE the gaming like Virtual Machines.

Despite Ryzen couldn't beat Intel, they did caught up and that is important... The competition is still alive.

 

[formulav8]

Despite Ryzen couldn't beat Intel

Actually Ryzen did in more than a few area's. AMD's IPC looks to be close to SL depending on app and Ryzen's SMT even looks to be better in area's. Again, looks to be certain scenario's as one would expect.

 

[aigomorla]

Actually Ryzen did in more than a few area's. AMD's IPC looks to be close to SL depending on app and Ryzen's SMT even looks to be better in area's. Again, looks to be certain scenario's as one would expect.

no...

cuz if we ignore budget and bring out intel's big guns, Ryzen gets spanked.
But that's IF we ignore budget.

For example u wont see comparisions to a 6950X because a overlocked 6950X would manhandle Ryzen.
But for fairness sakes the cpu alone would cost more then a fully optioned complete 1800X hence why we wont ever see a comparison.

 

[hackdrag0n]

no...

cuz if we ignore budget and bring out intel's big guns, Ryzen gets spanked.
But that's IF we ignore budget.

For example u wont see comparisions to a 6950X because a overlocked 6950X would manhandle Ryzen.
But for fairness sakes the cpu alone would cost more then a fully optioned 1800X hence why we wont ever see a comparison.

"Gets spanked" is probably exaggerating a little bit. Ryzen 1800x still looks respectable, performance wise, compared to a 6950x even if a bit behind.