Modern CPUs and gaming - good benchmarking/read from Tech Report

[kalniel]

The results we've compiled confront a popular myth: that PC processors are now so fast that just about any CPU will suffice for today's games, especially since so many titles are console ports. I've said something to that effect myself more than once. But is it true? We now have the tools at our disposal to find out. You may be surprised by what we've discovered.

http://techreport.com/articles.x/23246

[CAT-THE-FIFTH]

Saw that yesterday,and it was an interesting read.

However,one thing though - all of the CPUs produce over 50FPS in the games tested with an HD7950 and even with the steady-state frame rate,at least 30FPS is achieved. This is the problem - most consoles games run at around 30FPS to 60FPS anyway,so you argue that any reasonably modern CPU is producing high enough framerates anyway,perhaps not for us computer enthusiasts though!!

[kalniel]

As the article says, it's not about the FPS.

[CAT-THE-FIFTH]

As the article says, it's not about the FPS.

I also referred to the latency tests,ie, what they call the steady-state frame rate,where they mentioned the AMD FX CPUs were closer to 30FPS.

The FX processors are cranking out 99% of the frames in 33 milliseconds or so, which translates to a steady rate of 30 FPS—much lower than even the slower Intel processors.

All but one of them render the great majority of frames in under 23 milliseconds, which translates to a steady-state frame rate of just under 50 FPS.

Looking at modern consoles a lot of games run at 30FPS anyway,which is fine for tens of millions of games worldwide,and that is a simple FPS measure too.

Putting the price of the CPUs to one side(many of the Intel CPUs are better value for gaming due to better measurable performance),it still means all of the CPUs tested would be fine in reality.

Computer enthusiasts like us,fret about such things though,although the Intel CPUs would have a longer lifespan for PC gaming than an FX one which, is the important take home message. Its a shame they did not test the socket 1155 Core i3 and Pentium CPUs though. Would have loved to see the results for my Core i3 2100.

[shaithis]

A popular myth? Hmm, I don't know about that....the guy who wrote the article obviously doesn't read Hexus........

[kalniel]

A popular myth? Hmm, I don't know about that....the guy who wrote the article obviously doesn't read Hexus........

Well I've posted here that any modern CPU is fast enough, and Cat's saying roughly the same thing above. So so far, that statement's accurate.

Now whether the review disproves it or not is a matter of interpretation and tolerance of above normal frame waits in some games (but not others).

[CAT-THE-FIFTH]

This is the thing though - what is the defintion of "fast enough" for most gamers??

I suppose the only real way to do this,is to do a large scale test with games run at different framerates and latency settings,to see how many people do actually notice differences. It would involve a lot of work though.

[Willzzz]

A flaw in the article is that they only test a small selection of really quite similar games.

It's worth pointing out that in many games there is precious little difference between the i5-760 and the 3770k but they should include some games where CPU power is a genuine issue, for example Starcraft 2:

http://www.anandtech.com/bench/CPU/129

[kalniel]

A flaw in the article is that they only test a small selection of really quite similar games.

But they all have very different CPU dependencies?

It's worth pointing out that in many games there is precious little difference between the i5-760 and the 3770k but they should include some games where CPU power is a genuine issue, for example Starcraft 2:

http://www.anandtech.com/bench/CPU/129

Or Skyrim, which they did include. They also test at gaming resolutions, unlike the anandtech bench.

[CAT-THE-FIFTH]

However,SC2 is not a first person shooter though,so I would suspect people would be more tolerant of lower framerates and high latencies though.

[kalniel]

However,SC2 is not a first person shooter though,so I would suspect people would be more tolerant of lower framerates and high latencies though.

Framerates.. yes, if consistent. Inconsistent FPS and worse, latency spikes, are a killer for such a high APM game though.

[Willzzz]

They also test at gaming resolutions, unlike the anandtech bench.

Anandtech do test at 1080p as well, in this test they purposefully test at a low resolution in order to highlight the effect of the CPU rather than the GPU.

SC2 requires fast reactions to be competitive which aren't possible at low framerates.

[CAT-THE-FIFTH]

Framerates.. yes, if consistent. Inconsistent FPS and worse, latency spikes, are a killer for such a high APM game though.

But to the same level as an FPS?? I don't know and I suspect it depends on the type of player though. If you are one of the people competing in the big SC2 tournaments,maybe you need to get a CPU which pushes higher framerates and lower latencies. However,what percentage of players is that though??

However, I know quite a few SC2 players myself who put a decent number of hours into the game since release(even online matches),and none of them have Core i5 CPUs.

[shaithis]

Well I've posted here that any modern CPU is fast enough, and Cat's saying roughly the same thing above. So so far, that statement's accurate.

And plenty of people here (like myself) who recommend Intel CPUs for gamers......because the AMD ones haven't cut the mustard in all games and settings for a while.
Hence a lot of the "which CPU?" questions normally get a response of "What games will you be playing?"

For me, saying any modern CPU is fast enough is utter tosh. For starters, you have things like Atoms.....and that's before you go onto the borderline CPUs like dual cores playing on 64 man BF3 servers, or AMD CPUs playing poor console ports......and this looks like yet another CPU comparison done of BF3 single player.......hardly a true reflection of what most BF3 players find!

But then again, I guess my definition of "fast enough" is different......for me it means "60FPS at all times you are not being held back by the GPU".

The whole 30FPS for console gamers....well AFAIC they can keep on putting up with it on their £200 systems......but when spending £500+ on a PC you expect better and therefore "fast enough" should reflect that.

[CAT-THE-FIFTH]

But then again, I guess my definition of "fast enough" is different......for me it means "60FPS at all times you are not GPU-limited".

The whole 30FPS for console gamers....well AFAIC they can keep on putting up with it on their £200 systems......but when spending £500+ on a PC you expect better and therefore "fast enough" should reflect that.

So if presented with 20 PCs with different framerates and latencies in the same game - how many could you rank without visual confirmation of framerates or latencies??

The same goes with things like higher tessellation levels as opposed to medium or lower levels - how easy would it be for you to tell during normal gameplay without taking screenshots??

The article talks about the myth of most modern CPUs being good enough for gaming - what is defined as good enough is subjective and by who?? Computer enthusiasts like us?? Pro game players?? Young people?? Old people??

Consoles though are a massive data set, and obviously it seems for enough people lower framerates and probably higher latencies are "good enough" otherwise consoles would not be selling at all.

[Willzzz]

Anyone with a relatively weak CPU will be playing at low settings in order to achieve acceptable response rates.

While professional players may not worry about eye candy (although some eye candy is actually very useful) those semi-serious players probably wouldn't mind high response rates AND pretty graphics.