*OFFICIAL* M17x Benchmark Thread - Part 3

[DR650SE]

Thanks for the log file DR650SE while running Vantage. There were a couple of times where there was a little bit of throttling with the multiplier bouncing back and forth a little between 26 and 27 but overall it was very good. Sometimes between tests the load will drop way down and that also causes the peak multiplier to drop down too. That's just SpeedStep doing its thing because the CPU is lightly loaded and is not a sign of throttling.

When the C0% / Load was hitting 80% to 90%, a Core i7-720QM or i7-820QM would be giving you zero turbo boost so the maximum multiplier would be only 12 for the 720 or 13 for the 820. Your i7-920XM is running at more than twice that speed so if anyone is wondering if buying an i7-920XM or i7-940XM CPU is money well spent I'd have to say, definitely.

Great results and no clock modulation going on. Dell has finally got their act together and built a great performing laptop.

Thanks, I'll see if I can get more. The thing that made me think it was throttling was during the first CPU test. It will run for the first 17-18 sec or so at 3000-3100 operations per second. Then in the last 10 is where my score drops significantly. The operations suddenly drop to about 15-1600 and shoot up to 24-2500 or so. It continues to bounce back and fourth. and once it's started that, it carries into the next test. But it is only done when I get over 95* which is most of the time. Would reducing the TDP and maintaining the multiplier give it less heat and still enough juice to maintain the multi without crashing?

Well I'm off to test out the new Beta drivers. Be back in a few hours.

[unclewebb]

When you reduce the ThrottleStop TDP setting, that should result in less turbo boost when fully loaded and a reduced multiplier.

I'm not seeing a huge amount of CPU throttling in the log file that would cause the operations per second to drop significantly like you are seeing when testing so I'm not sure yet what's going on. Maybe if you have a watch synced up with the time on your laptop you can write down the exact time you are seeing this drop off to compare that to the log file. You might also have to log with GPU-Z at the same time to cover all bases.

Also try running a RealTemp log. It reports the individual core temperatures. I'm not sure about the 95C number being reported by HWiNFO32 so I'd like to see a second opinion. There seems to be too big of a difference in the maximum core temperature between two cores that are less than a cm away from each other. It might be sensor error or reporting error or who knows.

[DR650SE]

I'll definately check it tommorrow as well as the log file. Is it possible that only one core turbos up and thus causes the increase in heat? I looked at CPUID HWMonitor and I meant to get a screen shot, but I think it was similiar, there was a differance in 1,3 and 2,4. I'll check tommrrow with a variety of programs running together as well as GPU-Z

[DR650SE]

On another note, I raised the bar in Vantage


Vantage: P18420
GPU: 17776
CPU: 20667



When I get home I'm going to try an incorporate some dry ice into the equation in order for better cooling on the heatsinks. Unfortunately I don't have access here and I think it'll lower the temps enough to prevent any throttling and I can incorporate a 22K CPU score into the run.

[unclewebb]

Usually when one core gets hot, it heats up the adjacent core to a similar amount so you don't see large temperature differences between cores. It's likely just Intel's crappy sensors. They make great CPUs but have been known to cheap out on the temperature sensors they use. With bench scores like that, who cares.

[DR650SE]

Yea, that sounds about right. To me it just didn't make sense that there could be such a large differant when heat transfer alone would make it much closer. I agree though, it's all about the score board and how far your willing to go.

[Mandrake]

On another note, I raised the bar in Vantage


Vantage: P18420
GPU: 17776
CPU: 20667



When I get home I'm going to try an incorporate some dry ice into the equation in order for better cooling on the heatsinks. Unfortunately I don't have access here and I think it'll lower the temps enough to prevent any throttling and I can incorporate a 22K CPU score into the run.

Nice run, those clocks are not even that high. Once you improve cooling try 915|1250.

[Mandrake]

I found the Core-i version of XTU, I doubt it would work with the M17x but worth a try....

RapidShare: 1-CLICK Web hosting - Easy Filehosting

[johnksss]

hold on there speed racer....lol
1: run real temp off to the right of the screen
2: run an aver over clock first
3: bench only vantage cpu test 1
4: open task manager and leave it on performance
5: check the temps and task manager after the run. making sure all 4 cores are maxed.
6: set a bit higher over clock and run it again

keep repeating till the throttle kicks in...then we can go from there as to what to do next.

[DR650SE]

Hmm, I'm unfamiliar with XTU. What exactly is it? As for the cooling, I'm running it in front of an A/C, and using cold compresses on the heatsinks with the keyboard attached only by the ribbon. Notebook shuts off when it's unplugged Maybe in a few months I'll throw dry ice on the heatsinks.

@John, I'llgive it a shot tommorrow and report back what I find.