Excel speed test comparison against an FX-8350

[Tsavo]

AMD M300 mobile CPU@ 2 GHz: 104.55

 

[Tsavo]

3570K@ 3.2GHz 39.44

Hinky arsed PC is what this is. Fought me every single step of the way building it. Even the chassis tried to f*** me.

 

[Saffron]

Results from my computer below. My entire computer is stock clock speeds.

Run#1: 33.75s
Run#2: 34.27s
Run#3: 33.94s

Average Time: 33.98s

 

[John Tauwhare]

The 3570K is generally more balanced (i.e. no major surprises) but short of having started this discussion before you bought your CPU it's hard to see what else you could have done.

Because specific usage is always important and while the AT review of CPUs includes Excel doing a Monte Carlo Simulation, those results would lead you to the wrong conclusion for your usage:

That shows Piledriver doing very well indeed (while using a fair bit more power mind) but it's obviously a totally different load to yours.

Yes. Sigh. I checked all the benchmarks I could find when making the initial CPU choice. Thinking about it now, a raw Monte Carlo Simulation is hardly a typical application of Excel if most of the work is just throwing up random numbers. Business applications are more likely to be mainly sequential calculations. My benchmark is typical of what I use Excel for, but on a smaller scale. Shaving a few seconds off a quick macro isn't meaningful but taking minutes off a long duration macro is worth having.

So, to get a meaningful performance improvement I've had to go Intel. I got a full refund on the 990FXA motherboard and the FX-8350 is on eBay.co.uk and has already reached GBP130. I paid GBP155 so the loss is small, and the entertainment value has been mega.

 

[John Tauwhare]

Thanks for the new posts today. Edit: chart update Feb 03, 2013 (v7):
Vectronic leads with 24.52.

 

[John Tauwhare]

Got my new i7-2700K to 5.1GHz at 1.448V. Stable in P95 SmallFFTs at 4.8GHz, 1.344V, 141Watts, 82C but not stable beyond. At 5.1GHz it will BSOD in P95 Blend after a few minutes (70C) but ok to run WEI (7.8 and 7.9s), Passmark, and doing the Excel benchmark in 24.86 seconds. What an awesome chip, I'm WELL chuffed with it.

 

[Vectronic]

Pfft, 24.86, I could beat that... in fact I already did on my first run @ 4.5GHz.

I can get mine to 5.0GHz, might try that later just for shizzles.

Sad how underrepresented AMD is though, sure 8350 is basically as high as it gets, but.... figured there were a bit more AMD's floating around the forums.

 

[John Tauwhare]

Pfft, 24.86, I could beat that... in fact I already did on my first run @ 4.5GHz.

I can get mine to 5.0GHz, might try that later just for shizzles.

Damn, you're right. Chart corrected (above). Can anyone do better?

 

[Makaveli]

Damn, you're right. Chart corrected (above). Can anyone do better?

The only way someone will beat that is have a higher clocked IVY chip.

 

[Vectronic]

I think MSO 2013, and Win8 has a fairly large effect, an any Ivy Bridge with a higher clock on those two should work.

But I wasn't really concerned about "winning", I was fine with second place since the slowest run is probably the better measure, more consistent.

 

[Smartazz]

27.46 seconds on a 4.5GHz 2500K. 82.38 seconds for all three loops.

 

[Tsavo]

I paid GBP155 so the loss is small, and the entertainment value has been mega.

That, and you've kept a small number of us busy for a small while, too.

 

[Vectronic]

Ok well, 5.0GHz doesn't seem to be happening... I tried for an hour to get it stable, by the end I was worse off than the beginning...lol... 4.8, even 4.9 is easy. I had it up to 5.0GHz about a week ago, don't remember the settings I used.

Was really hoping to beat 20 seconds, didn't happen... did beat my previous 4.5GHz though, but I've had weeks to optimize that. (don't update list for that)

RAM is all @ 2000 10-10-10-26-2 (2x4GB)
4.5GHz: 24.33
4.6GHz: 23.69
4.8GHz: 22.87

Same as before, Windows 8, MSO 2013... 3570K.

 

[biostud]

35.79s i5-750 @ 3.8Ghz, win8 MSO 2010

 

[John Tauwhare]

Ok well, 5.0GHz doesn't seem to be happening... I tried for an hour to get it stable, by the end I was worse off than the beginning...lol... 4.8, even 4.9 is easy. I had it up to 5.0GHz about a week ago, don't remember the settings I used.

Was really hoping to beat 20 seconds, didn't happen... did beat my previous 4.5GHz though, but I've had weeks to optimize that. (don't update list for that)

RAM is all @ 2000 10-10-10-26-2 (2x4GB)
4.5GHz: 24.33
4.6GHz: 23.69
4.8GHz: 22.87

Same as before, Windows 8, MSO 2013... 3570K.

22.87! Congratulations. I can't get near that. Went up to 5.2 today at 1.472V but the best I could do was 25.04 so it's going backwards. Then at 4.8 I tried tweaking Excel and went into Options > Advanced > Formulas and switched from Use all processors: 8, to Manual 4 and got 24.76, my fastest time. Worth a try. I'll update the chart tomorrow.

 

[Vectronic]

That's a pretty useful tip really... 4.8GHz ends up being faster than your 5.2. Strange that it's plateauing so soon.

When this thread first started I tried all sorts of things to see if it had any effect, didn't seem to... RAM speed, SSD vs HD, Full window Vs. shrunken Vs minimized, no significant difference (might make a difference on older hardware though).

Taking a little data and stretching it... I get this:

So it looks like mine would plateau around the 5GHz mark too.

Edit: Tried exporting it as XLSM, and opening it in LibreOffice... it got like 25% through the first test, errored out because of some feature LO doesn't have regarding the statusbar.

Might try and do a conversion later. Nope, "fixed" the first error (commented it out, irrelevant value display)... 2 minutes later at about 50% through the first test... error because the table was smaller than the values it was looking for... screw it, takes too long to run it anyways...lol

 

[John Tauwhare]

I overcame my timidity and cranked up the volts from 1.472V to 1.496V at 5.2GHz and got a good run of 24.43, so no plateau yet.

Having discovered the benefit of setting Excel from 8 cores to 4, I ran the entire test again with HT off. I usually run 4 or 5 times hoping to get 3 consistent times together with a few outliers +~2secs which I assume is Windows doing something random in the background. With HT off not only is the run faster (-8% @4ghz, -6% @5GHz) but the good times are much tighter.

Best at 5.2 is 22.96 (HT off).

Will update the CPU/times chart next.

 

[John Tauwhare]

Chart updates Feb 04, 2013 (v8):

 

[John Tauwhare]

5.3GHz is next and could give 22.5. I'm not going for P95 stability with this, just boot and do a few runs, if it will do so with 1.5V ,as I'm not willing to go higher. Fortunately, temps during runs didn't exceed 60C at 5.2. There's not super-cooling in my case, just a fan at the front and one pulling through the heatsink:

 

[Face2Face]

John, looks like your are causing some trouble here in the CPU forum

I just bought a 3570K yesterday (On sale!) and I am itching to get this thing overclocked. I will post my scores when I get it up and going. BTW - I just picked up another Dell XPS Case, this time black with a 750watt PSU for $20. Spring project

Your new setup looks great BTW

 

[John Tauwhare]

.

John, looks like your are causing some trouble here in the CPU forum

I just bought a 3570K yesterday (On sale!) and I am itching to get this thing overclocked. I will post my scores when I get it up and going. BTW - I just picked up another Dell XPS Case, this time black with a 750watt PSU for $20. Spring project

Your new setup looks great BTW

Hey Face2Face. Just wanted to compare my FX-8350 doing Excel with Intel chips and it got shot down then things just got out of control... The mission crept so much I had to rebuild after a couple of weeks. An XPS case for $20 is a great deal - looking forward to see what you do with it. Good luck with the 3570K.

 

[guskline]

Funny how we get caught up in the OCing cycle

 

[Vectronic]

Oh come on, I've had mine up to 1.63v now, and it's an Ivy Bridge. (that's what, 0.11 over Intels supposed recommended max?) You have to keep going till you beat me again.

 

[Rvenger]

Oh come on, I've had mine up to 1.63v now, and it's an Ivy Bridge. (that's what, 0.11 over Intels supposed recommended max?) You have to keep going till you beat me again.

Some chips can take a beating like that, some can't. All part of the silicon lottery.

 

[Abwx]

The upsetting thing is that 99% of any chip die will take that but there s
1% that will fry.....