|
Hi. Are there some Ryzen 7 CPU players? Could You share to us some tests ? low/avg fps for example on test unit map with huge amount of units. We will be very much appreciated. I consider to buy Ryzen 1700 and I dont want to loose any current fps (i7-2600K)
Thanks, Cheers!
|
I wouldn't mind knowing FPS during an average game.
My 4720hq gets some ~200 at the start of a game on low, and somewhere below 60 during the late game. Something like that.
|
Thank You for reply. My question and this topic are stricte about Ryzen CPU handling Starcraft 2
|
sc2 only uses 2 cores, someone tested it with a strong dual core and overclocked it to 4 ghz and had the same fps as the top of i7 processors
the ryzen will suck for sc2
|
@Saggymidgetbooty6969, I know they are not monster of performance, but are they better or equal than my i72600K@4.4 ? I got some second hand test results and it seems like could have similar performance.
|
Dominican Republic587 Posts
just get the Ryzen 5 1600X is way better than a 2600x
|
Seeker
Where dat snitch at?36645 Posts
|
SC2 community, players and stuff stricte are interested to this page I think (http://www.teamliquid.net/forum/?filter=starcraft2) not main page forum (http://www.teamliquid.net/forum/)
...where are many subjects:
- SC2 General - StarCraft 2 Tournaments - StarCraft 2 Strategy - SC2 Maps & Custom Games - BW General - Brood War Tournaments - Brood War Strategy - User Streams - Tech Support - TL Community - General Forum - Other Games - Counter-Strike: Global Offensive - Heroes of the Storm - Super Smash Brothers - Sports - Media & Entertainment - TL Mafia - Blogs - Fan Clubs - Website Feedback - Legacy of the Void - Single Player - Overwatch
so.....this thread will die quickly here ⊙﹏⊙ Please consider to put this on SC2 General. It's match.
PS. On internet there are not any information or tests about Ryzen 7 and Starcraft 2. Zero opinions. Let's find out here on teamliquid first, straight from our friendly players シ
Cheers, Have a nice day
|
sc2 only uses two physical cores, maybe get something with better single core performance if you don't plan to do work like rendering.
|
This review in German has SC2 benchmarks on page 2, and something about better RAM on page 3:
http://www.pcgameshardware.de/Ryzen-5-1600X-CPU-265842/Tests/R5-1500X-Review-Mainstream-1225280/
They seem to be using some replay with a lot of units and explosions on screens, and at a very low resolution, so should be a good benchmark.
The raw architecture seems to compete well with Intel's, but the problem is clock speeds. Ryzen loses really bad if you include those newest Intel models that are using higher clock speeds by default. Ryzen would also lose really bad against the previous Intel models if you do manual overclocking on those. It seems Intel's stuff can be pushed a lot higher than Ryzen, and then benchmarks should look really bad if it's for example 4.8GHz on Intel's newer stuff compared to 4.0GHz on AMD Ryzen.
On page three of the review, there's faster 3200 RAM instead of 2667 adding 7% to FPS.
|
On April 15 2017 07:54 Saggymidgetbooty6969 wrote: sc2 only uses 2 cores, someone tested it with a strong dual core and overclocked it to 4 ghz and had the same fps as the top of i7 processors
the ryzen will suck for sc2
For the main part of the game SC2 is actually single-threaded, although it does use a second core for some things (from memory it's stuff like sound and AI). Which means the extra cores of Ryzen won't be of any benefit whatsoever for SC2.
That said, it won't *suck* because for single-threaded things Ryzen is quite fast. It's *not* as fast as Intel's newest processors, but it *is* faster than Sandy Bridge/Ivy Bridge CPUs (which is what I'm running) and those are often averaging >100fps,
|
|
Dominican Republic587 Posts
ryzen 5 has more thread more cores and on top of that has a really god IPC and better power consumption / temps than intel does, Ryzen is a platform that can last you longer than intel since you dont need to change your motherboard to upgrade your CPU.
|
United Kingdom20154 Posts
On April 15 2017 12:47 Ropid wrote:This review in German has SC2 benchmarks on page 2, and something about better RAM on page 3: http://www.pcgameshardware.de/Ryzen-5-1600X-CPU-265842/Tests/R5-1500X-Review-Mainstream-1225280/They seem to be using some replay with a lot of units and explosions on screens, and at a very low resolution, so should be a good benchmark. The raw architecture seems to compete well with Intel's, but the problem is clock speeds. Ryzen loses really bad if you include those newest Intel models that are using higher clock speeds by default. Ryzen would also lose really bad against the previous Intel models if you do manual overclocking on those. It seems Intel's stuff can be pushed a lot higher than Ryzen, and then benchmarks should look really bad if it's for example 4.8GHz on Intel's newer stuff compared to 4.0GHz on AMD Ryzen. On page three of the review, there's faster 3200 RAM instead of 2667 adding 7% to FPS.
SC2 benchmarks at low FPS are really weird because of engine quirks
You can have one CPU that's 20% faster than another (say if you tested a 7700k at 4.0 vs 4.8ghz) getting a 40% higher FPS number (average and minimum) while only perceptively being 20% faster.
That's because the FPS number even for minimum FPS is just a count of how many frames came in the last second but sc2 has extremely uneven frametimes with some frames taking 2-3x as long as others. A slower CPU might get stuck dealing with the game logic 90% of the time while a slightly faster one gets it done with say 75% and then gets a load of easy frames to render between simulation ticks; they massively inflate the FPS number without the actual part of the game that feels bad (the slow frames during simulation ticks) being improved at all.
That test seems to be scaling in that way because of the performance gap between CPU's and the minimum FPS range (below 20, sometimes far below) - sc2 runs at ~22hz, so that effect can get quite pronounced when FPS is dropping below the 40's. You can see that the ryzen cpu's are roughly level with the 6800.
I do expect Ryzen performance to be somewhere around Sandy Bridge for sc2 (OC vs OC)
|
I do expect Ryzen performance to be somewhere around Sandy Bridge for sc2 (OC vs OC)
Looks like could be true after my case but we should check by more tests. All I try to confirm is Ryzen 1700 lets to play comfortably or not.
Someone has Ryzen CPU? : )
|
Do you want to have it tested on highest settings? I'm using Ryzen 1700 (stock) and Geforce GTX 1070 with 16 gigabyte of RAM at 2133mhz because I didn't bother to configure XMP profiles yet. (I don't know if its single or dual rank RAM, but apparently it can make a difference of up to 10% for Ryzen CPUs in some games.)
|
United Kingdom20154 Posts
It's best to compare quite directly with two people repeating benchmarks from the same time of the same replay on the same settings, i usually used the fraps benchmark function to output frametimes as well as min,max,avg.
so.....this thread will die quickly here ⊙﹏⊙ Please consider to put this on SC2 General. It's match.
Yeah, that's unfortunate. Not neccesarily die but there is a lot less activity here since it got moved to be 2 and a half pages down the TL website a while back. There have been many website changes that were for the worse IMO, some with some positives and others that were just straight up bad but there is not much response to website feedback
|
I accidentally read over your entire replay post with the settings when I first read through the thread. For me the lowest FPS were 40 in the Zealot vs Ling/Bling fight. It wasn't a rapid drop in frame rate but rather continuous, so it didn't feel laggy at all for me. The other two fight's lowest FPS were 42(Voidray vs Corruptor) and 63 (Ultralisk vs Zealot). While there was no fight the FPS were between 80 and 120 depending on where the camera was and the movement of the units.
|
United Kingdom20154 Posts
You should disable Vsync for playing sc2 and especially for benchmarking it
I ran through that replay with standard benchmarking stuff (matched settings, locked to player camera, a hatchery selected) and started a 20 second fraps benchmark from 4:41 ingame time on Faster.
With a 6700k @4.5ghz and 3200mhz c16 RAM i got 87fps min, 119 avg, 180 max
frametime distribution, every dot is a frame:
Not exactly as it appears to be from the FPS number alone. The slowest frames during the start of the battle are hanging around 1/40'th of a second (40fps equivelant) even though the FPS number never dropped below 87 because of the highly irregular frametimes - There were 87 frames in that second but some of them were far, far slower and others far, far faster, it's just a very basic average over a time period (1 whole second) that is entirely too long for judging smoothness when frametimes are not consistent.
--
With some downclocking of my own CPU i can clearly see a disconnect between reported framerate and actual percieved performance. The percieved performance is very strongly tied to the amount of time that the slowest frames take while having a bunch of extra very fast frames between the slow ones does not benefit the feeling very much.
With a +25% to CPU clock speed i'm able to see the slowest frames get ~1.25x faster, the percieved performance getting 1.25x better but meanwhile the framerate increases by +40-45% - not a good representation of percieved performance. That will happen with RAM as well, the FPS goes up far faster than the engine and percieved performance does.
There's a set amount of work to be done per second and after that work is done, the CPU is free to make lots of frames that take much less time each. Those frames give a +1 to the FPS number but they don't speed up the game simulation and they don't really matter for percieved game performance so it makes things tricky to benchmark accurately.
Feels like i'm repeating myself a bunch and still not explaining adequately x.o
best bets, i think, would be running an extremely intense benchmark where Ryzen would get a minimum FPS in the low teens and even a 7700k would not be able to achieve >20 on (800 zerglings? ). Second best, run FRAPS benchmark set to record frametimes and then manually count out low long the slowest frames of the same fight took by comparison, that's what i did w/ different CPU clocks.
|
Canada13372 Posts
So... for game performance, Intel is still better than Ryzen?
I've seen people rave about ryzen for video encoding though.
So if I want to upgrade my PC in the future, I should stick on the intel OC train? Or are they close enough in 3D gaming performance that its worth saving the money and going ryzen?
|
United Kingdom20154 Posts
saving the money
The 7700k and r7 1700 cost the same, if you're talking about midrange stuff then it's more complicated but the 6c12t ryzens are great all around CPU's
|
Canada16217 Posts
On April 21 2017 22:42 ZeromuS wrote: So... for game performance, Intel is still better than Ryzen?
I've seen people rave about ryzen for video encoding though.
So if I want to upgrade my PC in the future, I should stick on the intel OC train? Or are they close enough in 3D gaming performance that its worth saving the money and going ryzen? While there is a difference between fps unless you only game, and want to spend more for a 7700k to get maximum frames than your best bet is the 1600. At this point getting an i5 is silly with the extra cores you get with Ryzen. Also when you're getting over 100+ frames losing a few isn't too much of a big deal, and if you play on a higher resolution the gap in performance lessens cause the gpu is doing more work.
|
United Kingdom20154 Posts
At this point getting an i5 is silly with the extra cores you get with Ryzen
It still depends on where your priorities are, a $65 pentium will beat any Ryzen CPU for sc2/WoW FPS etc. It's not just flat out better in every way but there's a large MT performance gain and relatively small ST loss when you're comparing OC'd Ryzen (6c12t+) to stock Kaby (non-7700k)
if you play on a higher resolution the gap in performance lessens cause the gpu is doing more work
That's only true when the GPU is capable of even lower performance than one or both CPU's.
Quick example: You might see two different CPU's get 45fps vs 60fps on 1080p, but when you run at 4k the graphics card will only do 20fps. You get 20fps on either system now.
That has not improved your 45fps at all, it's actually just made the game performance worse. When the game is running slower than you want it to run because of CPU performance, the last thing that anyone generally wants to do is to make it run even worse by making another part of the system weaker than the CPU that isn't giving the desired performance.
When you upgrade to a faster graphics card, you're now running 45fps vs 60fps again on 4k this time, hitting the limits of both CPU's again - the resolution hasn't magically made either of them faster
|
France12463 Posts
So basically when I play sc2 with my 7700k I can be sure my opponent will have an harder time playing the game (except if he has an OC'ed kaby)?
|
United Kingdom20154 Posts
Yeah, a 120hz+ gsync monitor is also a big thing for sc2
|
Canada13372 Posts
On April 22 2017 06:07 Cyro wrote:Show nested quote +At this point getting an i5 is silly with the extra cores you get with Ryzen It still depends on where your priorities are, a $65 pentium will beat any Ryzen CPU for sc2/WoW FPS etc. It's not just flat out better in every way but there's a large MT performance gain and relatively small ST loss when you're comparing OC'd Ryzen (6c12t+) to stock Kaby (non-7700k) Show nested quote +if you play on a higher resolution the gap in performance lessens cause the gpu is doing more work That's only true when the GPU is capable of even lower performance than one or both CPU's. Quick example: You might see two different CPU's get 45fps vs 60fps on 1080p, but when you run at 4k the graphics card will only do 20fps. You get 20fps on either system now. That has not improved your 45fps at all, it's actually just made the game performance worse. When the game is running slower than you want it to run because of CPU performance, the last thing that anyone generally wants to do is to make it run even worse by making another part of the system weaker than the CPU that isn't giving the desired performance. When you upgrade to a faster graphics card, you're now running 45fps vs 60fps again on 4k this time, hitting the limits of both CPU's again - the resolution hasn't magically made either of them faster
I was thinking of getting an i7 next as I have an i5 atm.
So if I was gonna get an i7 im better off with the higher end ryzen cpus in terms of maximizing my money? I do like to stream from time to time, and I currently use my laptop with a 3 series mobile i7 as my encoding machine. I could use my i5 machine for streaming and use the new processor for gaming.
I play a lot of overwatch and still some SC2 now, so I'm less concerned about pure single core performance vs before. If this helps provide some context.
|
United Kingdom20154 Posts
I was thinking of getting an i7 next as I have an i5 atm.
Which ones?
So if I was gonna get an i7 im better off with the higher end ryzen cpus in terms of maximizing my money?
For workloads that can utilize 16 threads efficiently, yes. For a general gaming system no.. or maybe
The 1700x/1800x are also not worth paying the 21-52% more over the 1700 because they're basically the same thing
|
Processors i5, i7 or Ryzen released in last 2 years do not constitute a bottleneck for sc2. They are all too fast for that. However, if you use any other software at the same time (like web browsers, especially chrome), skype, twitter etc, etc and especially if you stream and other background tasks at the same time then you should consider if you might need a high performance processor for your needs related to sc2. Otherwise performance issues will be related to memory, graphic card, network settings or other settings.
|
Canada13372 Posts
On April 24 2017 10:43 Cyro wrote:Which ones? Show nested quote +So if I was gonna get an i7 im better off with the higher end ryzen cpus in terms of maximizing my money? For workloads that can utilize 16 threads efficiently, yes. For a general gaming system no.. or maybe The 1700x/1800x are also not worth paying the 21-52% more over the 1700 because they're basically the same thing
I was going to get a 6700K or 7700K Overclockable i7 that fits into a price range I can afford with a motherboard.
I do like to dabble in video editing though, and I've been trying it out some more lately so it would be nice to have a computer that handles it a bit more efficiently than my current, older i5.
For gaming at 1080 I know I dont NEED more than an i5, but if I am going to be getting a new motherboard and processor anyway, I can stretch my money further over time and have more capabilities if I go the i7 route no?
|
United Kingdom20154 Posts
Processors i5, i7 or Ryzen released in last 2 years do not constitute a bottleneck for sc2. They are all too fast for that.
The fastest systems on the highest FPS settings cannot keep all frames above 1/60'th of a second in a high or sometimes even moderately high supply 1v1 battle. There are also large performance differences between stuff like Kaby and the earlier intel CPU's / Ryzen and from overclock vs stock, fast RAM vs slow RAM etc - so no, not fast enough :D
------
but if I am going to be getting a new motherboard and processor anyway, I can stretch my money further over time and have more capabilities if I go the i7 route no?
Yeah, 7700k is basically just a better 7600k. No tradeoffs there
|
The fastest systems on the highest FPS settings cannot keep all frames above 1/60'th of a second in a high or sometimes even moderately high supply 1v1 battle. There are also large performance differences between stuff like Kaby and the earlier intel CPU's / Ryzen and from overclock vs stock, fast RAM vs slow RAM etc - so no, not fast enough :D
As I said, its not CPU that is the bottleneck in sc2. Each will give you over 120 f/s. The 60 f/s is the g-sync bug. Low frame rates do not result from processor speed but from software used, especially malicious software and various settings and other hardware.
As to the 7700k and 6700k processors, they are just about the same. Unless you think you will notice a 2-3% difference that will be dilluted by the rest of your hardware.
|
United Kingdom20154 Posts
As I said, its not CPU that is the bottleneck in sc2. Each will give you over 120 f/s
No, they won't. The CPU load is proportional to the amount of action, you can have 500fps in the first moments of a game and then effectively sub-60 during a high supply fight
Increasing/decreasing CPU speed alone will change performance while other parts of the system such as graphics performance will not, hence CPU bottleneck
The 60 f/s is the g-sync bug
It's not related to g-sync at all
As to the 7700k and 6700k processors, they are just about the same
about +300mhz (~6.4%) so kinda, but a little better
|
France12463 Posts
On April 24 2017 23:55 Kafka777 wrote: Processors i5, i7 or Ryzen released in last 2 years do not constitute a bottleneck for sc2. They are all too fast for that. However, if you use any other software at the same time (like web browsers, especially chrome), skype, twitter etc, etc and especially if you stream and other background tasks at the same time then you should consider if you might need a high performance processor for your needs related to sc2. Otherwise performance issues will be related to memory, graphic card, network settings or other settings. I don't think so. Even tho I almost always play with Chrome open, I don't think that my playing conditions are always perfect, despite playing on a (albeit stock) 7700k with fast RAM (3200MHz). When I play a TvZ and we both reach 180+ supply with the zerg having a lot of lings, like 100+, it's not as easy to micro, so I am a bit salty that it will primarily affect me since I'm the T and I risk getting fungal'ed and stuff. (144Hz monitor) So I don't even want to imagine what it would feel like on Ryzen. edit: and I play with physics disabled, etc... every thing where CPU performance is affected I lower or disable.
|
about +300mhz (~6.4%) so kinda, but a little better
It has higher clock and lower power consumption - at the cost of efficiency. So no, its not 6.4% faster.
I don't think so. Even tho I almost always play with Chrome open, I don't think that my playing conditions are always perfect, despite playing on a (albeit stock) 7700k with fast RAM (3200MHz). When I play a TvZ and we both reach 180+ supply with the zerg having a lot of lings, like 100+, it's not as easy to micro, so I am a bit salty that it will primarily affect me since I'm the T and I risk getting fungal'ed and stuff. (144Hz monitor) So I don't even want to imagine what it would feel like on Ryzen. edit: and I play with physics disabled, etc... every thing where CPU performance is affected I lower or disable
You will notice the 7700k cannot handle memory faster than 2400 unless overclocked? You realize 3200 memory has higher latency than 2400 so in fact it is slower? Anyway - this will not make any difference really, if you have at least 16 gb ram. I use a lot of PC's and upgrade one at a time, my kids usually run into low frame rate problems with games, including sc2, which will run very well on an old 3 core sub 3 ghz processor. The only real problem I noticed is the way they use software - this has a significant impact - especially chrome browser - for whatever reason. They use up too much resources for other things. Some new software also takes up too much resources even if apparently it is not used, but works in background. Getting much better cpu helps, but it will not be as helpful as understanding hardware and software.
|
France12463 Posts
On April 25 2017 06:41 Kafka777 wrote:It has higher clock and lower power consumption - at the cost of efficiency. So no, its not 6.4% faster. Show nested quote +I don't think so. Even tho I almost always play with Chrome open, I don't think that my playing conditions are always perfect, despite playing on a (albeit stock) 7700k with fast RAM (3200MHz). When I play a TvZ and we both reach 180+ supply with the zerg having a lot of lings, like 100+, it's not as easy to micro, so I am a bit salty that it will primarily affect me since I'm the T and I risk getting fungal'ed and stuff. (144Hz monitor) So I don't even want to imagine what it would feel like on Ryzen. edit: and I play with physics disabled, etc... every thing where CPU performance is affected I lower or disable You will notice the 7700k cannot handle memory faster than 2400 unless overclocked? You realize 3200 memory has higher latency than 2400 so in fact it is slower? . It's not as simple as that, the 3200MHz memory is still giving better performances than most (or all?) 2400 ones. As for the 7700k handling faster memory, I hope you do realize it's just ticking a box in the BIOS in order to activate XMP profiles? My software is fine by the way.
|
United Kingdom20154 Posts
It has higher clock and lower power consumption - at the cost of efficiency. So no, its not 6.4% faster.
The performance per clock is the same so 6.4% higher clocks is 6.4% more performance
-----
You realize 3200 memory has higher latency than 2400 so in fact it is slower?
2400/15 = 160 3200/16 = 200
16 cycles at 3200mhz takes ~1.25x less time than 15 cycles at 2400mhz - that's the latency
3200/2400 = 1.33x which is the bandwidth improvement in a sustained transfer
The RAM performance does impact sc2 framerate
The game does run okay on lots of hardware, it just doesn't run great/amazing on anything and there is a wide range of performance between different cpu's
|
It's not as simple as that, the 3200MHz memory is still giving better performances than most (or all?) 2400 ones. As for the 7700k handling faster memory, I hope you do realize it's just ticking a box in the BIOS in order to activate XMP profiles? My software is fine by the way.
The processor is simply unable to handle faster memory - this is a known bottleneck. It does not matter if you activate XMP. 7700k is only able to handle 2400 according to specification (in turbo mode). If specification is true (and it is close enough) then you will only be able to utilize 3200 memory specifications if you overclock your processor. However, as I mentioned, you will not lose out much on this. Ram, speeds are about the same all around in real usage simply because the processor is the bottleneck. Now what they give you from all this advertising - journalist pages is just crap.
|
France12463 Posts
On April 25 2017 07:10 Kafka777 wrote:Show nested quote +It's not as simple as that, the 3200MHz memory is still giving better performances than most (or all?) 2400 ones. As for the 7700k handling faster memory, I hope you do realize it's just ticking a box in the BIOS in order to activate XMP profiles? My software is fine by the way. The processor is simply unable to handle faster memory - this is a known bottleneck. It does not matter if you activate XMP. 7700k is only able to handle 2400 according to specification (in turbo mode). If specification is true (and it is close enough) then you will only be able to utilize 3200 memory specifications if you overclock your processor. However, as I mentioned, you will not lose out much on this. Ram, speeds are about the same all around in real usage simply because the processor is the bottleneck. Now what they give you from all this advertising - journalist pages is just crap. Why do they show 1600 MHz on CPU-Z then?
|
hy do they show 1600 MHz on CPU-Z then?
I have no idea what you are referring to exactly.
|
Canada13372 Posts
So, I was checking prices and the 7700k is 20 dollars more in CAD than the Ryzen 1700 so I may as well take the unlocked processor.
I'll keep an eye out for when a mobo bundle ends up existing :D Only need 1 gfx card slot, don't see myself doing 2 cards anytime soon.
Thanks for the help friends.
Or I could get a 6700k with mobo for only 620 hrmmmmm
|
Maybe some ryzen's owners could share to us their experience on playing. These are only speculations : )
|
So, I was checking prices and the 7700k is 20 dollars more in CAD than the Ryzen 1700 so I may as well take the unlocked processor.
I'll keep an eye out for when a mobo bundle ends up existing :D Only need 1 gfx card slot, don't see myself doing 2 cards anytime soon.
Thanks for the help friends.
Or I could get a 6700k with mobo for only 620 hrmmmmm
While the difference between 6700k and 7700k is negligable, the 7700k is nicer to have, the difference between 7700k and Ryzen 1700 is not so much in overclocking potential of 7700k but in number of cores/thread (4/8 vs 8/16) for most current PC games 7700k will probably be better due to higher frequency that is unobtainable for cpu's with more cores, for VR or almost anything else the Ryzen will be better and is more future proof. Ryzen 1700 is at least 20% slower in single core (first core) performance (which is important for current games) but hugely more powerful in multitasking (its a question whether it can be utilized properly in standard home usage). The Ryzen input is that multicore processors will become popular faster and the people creating software will have much more options available. Nevertheless these processors will be utilized properly by masses in years to come and only by few right now - just like Intel extreme cpu's.
|
United Kingdom20154 Posts
the 1700 does have a massive MT lead but the core performance is lower (~1.35x lower than a 7700k, OC vs OC) and the amount of threads that you have to efficiently scale to in order to overcome that and then get a lot better performance is very high (8-16), way too high for almost all games now and in the next few years
You can see this with most games performing almost the same on an ryzen 1600 vs 1700. Going from 6c12t to 8c16t is a +33.3% increase to core count but often only +0-5% in actual performance because of how underutilized they are
|
|
|
|