CPUs compared clock for clock. FSXmarkCPU

[SAAB340 28]

 

 

Can someone please tell me the correct ZOOM for this test? I am doing it but since a newly created FSX.CFG defaults the "wideviewaspect=" line to false, what is the right zoom? It appears when loading the flight it is .70 Is this right?

Yes, the default zoom of 0.70 is the correct one. What CPU are you using? Thanks for testing.

[cvearl 60]

Yes, the default zoom of 0.70 is the correct one. What CPU are you using? Thanks for testing.

Hey cool. Thanks. You won't want my results. I use an AMD gpu. Wanted to check if my cpu was in line with expectations. 2600k on P8P67.

 

C.

[cvearl 60]

OK FSX.cfg and ZOOM again... Here's my problem. I understand .70 is right for ZOOM. Following your instructions to build a fresh default FSX.CFG, wideviewaspect=false which renders a much higher score than wideviewaspect=true and false looks weird.

 

Examples... (these are both paused so the fps are higher than the test would be). So which is valid to run the test?

 

 

PICTURE 1 = FSX default CFG (Wideviewaspect=false)

 

 

 

 

FSX.CFG modified with wideviewaspect=true. Lower framerate as a result as well.

 

 

[SAAB340 28]

OK FSX.cfg and ZOOM again... Here's my problem. I understand .70 is right for ZOOM. Following your instructions to build a fresh default FSX.CFG, wideviewaspect=false which renders a much higher score than wideviewaspect=true and false looks weird.

 

Examples... (these are both paused so the fps are higher than the test would be). So which is valid to run the test?

 

 

PICTURE 1 = FSX default CFG (Wideviewaspect=false)

 

 

 

 

FSX.CFG modified with wideviewaspect=true. Lower framerate as a result as well.

 

 

 

The top picture looks exactly as it should. Default cfg with only AM and BP entry added. No tinkering with anything in the default cfg.

[cvearl 60]

The top picture looks exactly as it should. Default cfg with only AM and BP entry added. No tinkering with anything in the default cfg.

Good to know. Thanks. I'll test.

[cvearl 60]

Ok I ran a clean setup by the book here. It was easier than I thought. AMD GPU (HD 7970 Ghz 3GB) so this is just academic and not official. Detail files enroute to you now Lars.

 

1. Ran REX and restored default textures.

2. Ran FTX Central and set to FSX DEFAULT

3. Deleted FSX.CFG and rebuilt and said NO to all the trusted DLL's that asked to run.

4. Flushed Shader Cache

 

For all intents and purposes a naked FSX install. Followed everything else to the letter.

 

I averaged 39.35 fps.

 

 2600K @ 4.0 (40x100) and DDR 1600 XMP profile same timings as you requested. Pretty spot on result to what you show on the chart. Looks like I'm going haswell at some point soon.

 

Charles.

[SAAB340 28]

Ok I ran a clean setup by the book here. It was easier than I thought. AMD GPU (HD 7970 Ghz 3GB) so this is just academic and not official. Detail files enroute to you now Lars.

 

1. Ran REX and restored default textures.

2. Ran FTX Central and set to FSX DEFAULT

3. Deleted FSX.CFG and rebuilt and said NO to all the trusted DLL's that asked to run.

4. Flushed Shader Cache

 

For all intents and purposes a naked FSX install. Followed everything else to the letter.

 

I averaged 39.35 fps.

 

2600K @ 4.0 (40x100) and DDR 1600 XMP profile same timings as you requested. Pretty spot on result to what you show on the chart. Looks like I'm going haswell at some point soon.

 

Charles.

Thanks for doing it Charles. I've recieved the result files but won't be able to look at them until next week as I'm away at the moment.

[SAAB340 28]

The i7 SandyBridge (2600K) has been added.

 

I've looked at Charles result files and they are all 'normal'. It is possible to run this benchmark with an AMD GPU. You just have to make sure you have a clean install. That's the same for nVidia users as well.

 

Regular SandyBridge performs within 1% of SandyBridge-E. The extra 4MB L3 on the 3930K doesn't make its IPC faster than the 2700K. The L3 cache on SandyBridge-E is larger, but at the same time it's also higher latency than in the regular SandyBridge. End result, pretty much a tie in FSX.

 

The result chart is slowly getting filled. It would be interesting with Haswell i5, SandyBridge i5 and also results from all the AMD CPUs.

 

Thanx for running the test cvearl.

[cvearl 60]

The i7 SandyBridge (2600K) has been added.

I've looked at Charles result files and they are all 'normal'. It is possible to run this benchmark with an AMD GPU. You just have to make sure you have a clean install. That's the same for nVidia users as well.

Regular SandyBridge performs within 1% of SandyBridge-E. The extra 4MB L3 on the 3930K doesn't make its IPC faster than the 2700K. The L3 cache on SandyBridge-E is larger, but at the same time it's also higher latency than in the regular SandyBridge. End result, pretty much a tie in FSX.

The result chart is slowly getting filled. It would be interesting with Haswell i5, SandyBridge i5 and also results from all the AMD CPUs.

Thanx for running the test cvearl.

No prob!

[ArjenVdv 190]

Wait, there is a difference between i5 and i7?

I always told that cache size was beneficial. By the looks of it, it is.

[SAAB340 28]

Wait, there is a difference between i5 and i7?

 

I always told that cache size was beneficial. By the looks of it, it is.

That's what the data says. Anyone else happy to run the benchmark is more than welcome to attribute, even if it's just a rerun to create another data point. I'd love to see an AMD CPU in here thou. Anyone?

[RobbieHe 0]

This is really good information. Thanks for the hard work putting it together. Hope we see more of this in the future!

 

RH

[Jettix 1]

Hello

 

4670k results here, VGA 560ti (not oc'ed)

 

Frames: 11083 - Time: 240000ms - Avg: 46.179 - Min: 30 - Max: 66

Frames: 11217 - Time: 240000ms - Avg: 46.738 - Min: 30 - Max: 67

Frames: 11208 - Time: 240000ms - Avg: 46.700 - Min: 29 - Max: 67

Frames: 10930 - Time: 240000ms - Avg: 45.542 - Min: 28 - Max: 65

Frames: 10927 - Time: 240000ms - Avg: 45.529 - Min: 28 - Max: 66

Frames: 11139 - Time: 240000ms - Avg: 46.413 - Min: 30 - Max: 66

 

SAAB340 How can I send you fraps' log?

[SAAB340 28]

Hello

 

4670k results here, VGA 560ti (not oc'ed)

 

Frames: 11083 - Time: 240000ms - Avg: 46.179 - Min: 30 - Max: 66

Frames: 11217 - Time: 240000ms - Avg: 46.738 - Min: 30 - Max: 67

Frames: 11208 - Time: 240000ms - Avg: 46.700 - Min: 29 - Max: 67

Frames: 10930 - Time: 240000ms - Avg: 45.542 - Min: 28 - Max: 65

Frames: 10927 - Time: 240000ms - Avg: 45.529 - Min: 28 - Max: 66

Frames: 11139 - Time: 240000ms - Avg: 46.413 - Min: 30 - Max: 66

 

SAAB340 How can I send you fraps' log?

You've got PM=)

[SAAB340 28]

I've had a look at the Haswell i5 results.

 

This is what it looks like if I plot the FPS averaged from all the 5 final benchmarks from Haswell & IvyBridge i5 & i7.

 

 

If we look at the Haswell i5 Vs i7 first there are a few places where the i5 actually is slightly faster than the i7 on average. But there is a variance between the benchmarks in those places if you were to check them individually. Especially the i7 results show a greater spread between benchmarks. I don't know if it could be due to very early BIOSes in the i7 results, or something else on the i7 system that is causing the variance. I'd say that the difference between i5 and i7 is within the margin of error as we are looking at different systems. (The total difference here is 1.6%)

 

When we look at the IvyBridge i5 vs i7 results they show a very different picture where the i5 is consistently a lot slower than the i7. I really wonder why that happens to the IvyBridge and not the Haswell?!?.

I wonder if the IvyBridge i5 results are running on a CPU set at stock speed? Because if it was set at stock speeds this it would explain the difference.

 

I've asked the question but still awaiting an answer. Best thing would be if someone else with an i5 IvyBridge would do the test.

It would be very helpful if we got more data points for any CPU, even for the once we've already tested.

 

For now I've decided to remove the IvyBridge i5 result awaiting clarification, and the Haswell i5 has been added. Thanks Jettix for providing the data.