Tech Report - How CPUs Affect Gaming Performance

***MayImilae*** · (This post was last modified: 08-25-2012, 10:03 AM by MayImilae.)

This is not entirely relevant for dolphin (hence why I put it in generally discussion) but it's still very interesting. Using their new frame latency scoring, tech report gives alot of interesting information. Take a look.

http://techreport.com/articles.x/23246/1

TL-DR - AMD chips suck even worse than we thought Tongue

.

**Anti-Ultimate** · 08-25-2012, 08:49 PM #2

(08-25-2012, 10:03 AM)MaJoR Wrote: This is not entirely relevant for dolphin (hence why I put it in generally discussion) but it's still very interesting. Using their new frame latency scoring, tech report gives alot of interesting information. Take a look.

http://techreport.com/articles.x/23246/1

TL-DR - AMD chips suck even worse than we thought .

Lol, AMD said that they don't want performance in their CPU's anymore Tongue

**AnyOldName3** · 08-25-2012, 09:06 PM #3

Yup, they're going for the low end market, which means that they'll only have a share of half the market, so will inevitably make smaller profits (although they won't spend so much on R+D).

WarWeeny · 08-26-2012, 12:39 AM #4

the next architechture of amd seems promising Tongue

linkinworm · 08-26-2012, 03:04 AM #5

so does haswell tbh, 10% faster clock for clock than ivybridge, but who knows skylake might see a better improvement after that and is only about 3 years away

**NaturalViolence** · (This post was last modified: 08-26-2012, 03:13 AM by NaturalViolence.)

MaJoR Wrote:Using their new frame latency scoring, tech report gives alot of interesting information. Take a look.

Frame latency is just the reciprocal of framerate. All of there information is derived from the framerate, they're just showing the data in different forms.

MaJoR Wrote:TL-DR - AMD chips suck even worse than we thought

Not really. There were countless gaming benchmarks on major hardware review sites many months ago when the FX series was first released that showed the exact same results.

Anti-Ultimate Wrote:Lol, AMD said that they don't want performance in their CPU's anymore

No they didn't. You're completely mixing up what they said.

AnyOldName3 Wrote:Yup, they're going for the low end market, which means that they'll only have a share of half the market, so will inevitably make smaller profits (although they won't spend so much on R+D).

What?

First off the "low end" (I prefer the term mobile) market is much bigger right now in terms of potential revenue than the desktop/server market which means if they successfully break into it there sales should increase not decrease. Second of all they're not pulling out of the desktop/server market completely they're just "shifting their focus" (by which they mean shifting their R & D budget and manufacturing) towards products intended for the mobile market.

WarWeeny Wrote:the next architechture of amd seems promising

That's what they said about the previous 3 architectures. I'll believe it when I see it.

WarWeeny · 08-26-2012, 04:02 AM #7

(08-26-2012, 03:10 AM)NaturalViolence Wrote:
WarWeeny Wrote:the next architechture of amd seems promising

That's what they said about the previous 3 architectures. I'll believe it when I see it.

true enough Tongue

***MayImilae*** · 08-26-2012, 10:42 AM #8

(08-26-2012, 03:10 AM)NaturalViolence Wrote:
MaJoR Wrote:TL-DR - AMD chips suck even worse than we thought

Not really. There were countless gaming benchmarks on major hardware review sites many months ago when the FX series was first released that showed the exact same results.

The new information I was referring to is how, on an FX CPU and a great GPU, the FX can create lag spikes and ruin the feeling of smoothness. Yet another architecture issue. I read several reviews of FX processors (fun read most of them) and most of them focused on benchmarks of the traditional sort, and didn't dive into how it would affect games like that.

**NaturalViolence** · (This post was last modified: 08-26-2012, 11:48 AM by NaturalViolence.)

Well it's a logical conclusion to draw based on the lower framerate. If a cpu receives a lower framerate in a benchmark you know that the lag spikes are going to worse.

Framerate in benchmarks is typically an average. Therefore if a particular cpu receives a lower framerate you can logically conclude that the minimum framerate, maximum framerate, or both are also lower (since that's the only way to bring the average framerate down). Since the GPU will limit the maximum framerate in most situations you can expect the bulk of that to be due to a decrease in minimum framerate.

If you think about all of the things that games do that are very cpu or memory heavy most of them are things that happen occasionally rather than constantly (physics calculations done by the physics engine for example). Therefore cpu performance really only usually effects minimum performance, reducing the severity of lag spikes, which is why their effect on average framerate tends to be low. It's been this way ever since GPUs started supporting hardware transform and lighting. Before that CPUs handled a lot of the 3D rendering work for video games with software rendering, which put a constant high load on the cpu (therefore you needed a good cpu to get a good framerate, period).