Check that the VCCSA and VCCIO are available to tweak. They're a bit fragile. There can be fractions of volts between recommended maximums and serious damage zones, but they've been known to provide some stability (particularly for the RAM).

The ASRock board looks okay. Power delivery looks beefy enough and decently cooled. ASUS boards are also a good pick.

Just be sure to give this some decent thought before going in. You don't want to be getting buyer's remorse... +$20 for an overclockable chip, +$50 for RAM that scales accordingly, $50 for a strong enough cooler, and you can get up to [Throw a flippin' rock] more performance out of Haswell.


EDIT: Scratch the $50 for the RAM since you're already going for 2400 MHz.

Not to be saying you're wrong in avoiding overclocking, but it's just so much more fun :smile:

For prep, I've bought [URL="http://www.newegg.com/Product/Product.aspx?Item=N82E16817139028"]this[/URL] PSU :smile: Very nice price.

I find it hilarious how PSU manufacturers and heatsink manufacturers still religiously advertise "Multi-Core READY!"

UPDATE: I converted one Haswell to Ubuntu 12.04 and the other to Windows 7. In 4 weeks, neither box has spontaneously rebooted.

This does not prove there is a bug in Windows 8. It could well be that the Windows 8 scheduler or Windows 8 device drivers do things in a different order such that a real hardware instability is revealed.

Is Haswell version of Prime95 available for download? I’d like to try it on a Lenovo Yoga 2 Pro, i5. Windows 8.1, 64-bit.

Thanks.

[QUOTE=TObject;362019]Is Haswell version of Prime95 available for download? I’d like to try it on a Lenovo Yoga 2 Pro, i5. Windows 8.1, 64-bit.

Thanks.[/QUOTE]

There are Windows and Linux versions attached to posts about 15 pages back. The Haswell version is 28.1.

[QUOTE=Mark Rose;362021]There are Windows and Linux versions attached to posts about 15 pages back. The Haswell version is 28.1.[/QUOTE]

Thank you, it is in the post [url=http://mersenneforum.org/showpost.php?p=350674&postcount=238]#238[/url]

Lenovo Yoga 2 Pro, 4GB/128GB/i5 Windows 8.1
Haswell, 22 nm, i5-4200U 1.6 GHz (2 GHz Boost)

32M exponent (FFT 1728K) -
1 LL thread : 0.012 sec per iteration
1 LL thread + 1 helper 0.007 sec per iteration

[code]Intel(R) Core(TM) i5-4200U CPU @ 1.60GHz
CPU speed: 2294.71 MHz, 2 hyperthreaded cores
CPU features: Prefetch, SSE, SSE2, SSE4, AVX, AVX2, FMA
L1 cache size: 32 KB
L2 cache size: 256 KB, L3 cache size: 3 MB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
TLBS: 64
Prime95 64-bit version 28.1, RdtscTiming=1
Best time for 768K FFT length: 8.783 ms., avg: 9.034 ms.
Best time for 896K FFT length: 10.738 ms., avg: 10.807 ms.
Best time for 1024K FFT length: 12.354 ms., avg: 12.468 ms.
Best time for 1280K FFT length: 14.219 ms., avg: 14.989 ms.
Best time for 1536K FFT length: 19.024 ms., avg: 19.080 ms.
Best time for 1792K FFT length: 21.640 ms., avg: 21.706 ms.
Best time for 2048K FFT length: 25.132 ms., avg: 25.222 ms.
Best time for 2560K FFT length: 31.348 ms., avg: 31.413 ms.
Best time for 3072K FFT length: 37.560 ms., avg: 37.621 ms.
Best time for 3584K FFT length: 45.562 ms., avg: 45.680 ms.
Best time for 4096K FFT length: 52.359 ms., avg: 52.437 ms.
Best time for 5120K FFT length: 63.341 ms., avg: 63.450 ms.
Best time for 6144K FFT length: 79.605 ms., avg: 79.736 ms.
Best time for 7168K FFT length: 91.420 ms., avg: 91.597 ms.
Best time for 8192K FFT length: 109.726 ms., avg: 110.459 ms.
Timing FFTs using 2 threads on 1 physical CPUs.
Best time for 768K FFT length: 3.319 ms., avg: 3.398 ms.
Best time for 896K FFT length: 3.985 ms., avg: 4.148 ms.
Best time for 1024K FFT length: 4.593 ms., avg: 4.702 ms.
Best time for 1280K FFT length: 5.658 ms., avg: 5.754 ms.
Best time for 1536K FFT length: 6.965 ms., avg: 7.143 ms.
Best time for 1792K FFT length: 8.229 ms., avg: 8.334 ms.
Best time for 2048K FFT length: 9.399 ms., avg: 9.755 ms.
Best time for 2560K FFT length: 11.830 ms., avg: 12.323 ms.
Best time for 3072K FFT length: 14.238 ms., avg: 14.540 ms.
Best time for 3584K FFT length: 17.102 ms., avg: 17.297 ms.
Best time for 4096K FFT length: 19.585 ms., avg: 19.853 ms.
Best time for 5120K FFT length: 24.165 ms., avg: 24.457 ms.
Best time for 6144K FFT length: 30.165 ms., avg: 30.660 ms.
Best time for 7168K FFT length: 34.877 ms., avg: 35.152 ms.
Best time for 8192K FFT length: 40.892 ms., avg: 41.314 ms.
Timing FFTs using 4 threads on 2 physical CPUs.
Best time for 768K FFT length: 2.657 ms., avg: 2.687 ms.
Best time for 896K FFT length: 3.213 ms., avg: 3.239 ms.
Best time for 1024K FFT length: 3.738 ms., avg: 3.761 ms.
Best time for 1280K FFT length: 4.832 ms., avg: 4.868 ms.
Best time for 1536K FFT length: 5.654 ms., avg: 5.907 ms.
Best time for 1792K FFT length: 7.156 ms., avg: 7.196 ms.
Best time for 2048K FFT length: 7.806 ms., avg: 7.871 ms.
Best time for 2560K FFT length: 9.894 ms., avg: 10.234 ms.
Best time for 3072K FFT length: 11.824 ms., avg: 12.004 ms.
Best time for 3584K FFT length: 14.080 ms., avg: 14.304 ms.
Best time for 4096K FFT length: 16.567 ms., avg: 16.668 ms.
Best time for 5120K FFT length: 21.152 ms., avg: 21.601 ms.
Best time for 6144K FFT length: 26.556 ms., avg: 26.905 ms.
Best time for 7168K FFT length: 30.821 ms., avg: 31.302 ms.
Best time for 8192K FFT length: 35.232 ms., avg: 35.565 ms.
Best time for 61 bit trial factors: 3.407 ms.
Best time for 62 bit trial factors: 3.490 ms.
Best time for 63 bit trial factors: 4.174 ms.
Best time for 64 bit trial factors: 4.415 ms.
Best time for 65 bit trial factors: 6.348 ms.
Best time for 66 bit trial factors: 7.703 ms.
Best time for 67 bit trial factors: 7.678 ms.
Best time for 75 bit trial factors: 7.517 ms.
Best time for 76 bit trial factors: 7.509 ms.
Best time for 77 bit trial factors: 7.492 ms.[/code]

Your one thread one CPU results look odd. Wasn't there something working in the background forna while?

[QUOTE=ldesnogu;362082]Your one thread one CPU results look odd. Wasn't there something working in the background forna while?[/QUOTE]

Not really. I just re-run the benchmark test, letting the Prime95 stabilize crunching a double-check for a while, beforehand. And I got similar results.

Not really... :razz: actually the first and the third test look ok, and they follow the theory about speed proportions. Is the second test the one which is odd. There you use the same core, but with a "logical" helper, so the results should be [U][B]closer to the first test[/B][/U], and not to the third. As it appears, your windoze 8 is using both physical cores there**, your windoze's 8 affinity scramble is... scrambled... FUBAR, etc, hehe... Not that it would matter, for "real life production", but just for the sake of. When you start crunching with 2 or 4 threads, it will not matter (I still recommend turning the HT off! but his is up to you - and [B][U]I might be wrong[/U][/B], I don't own a Haswell, neither played with one).


---------
** to be clear: that test, the second, is in reality done with "2 threads, on 2 physical CPU cores" due to the fact that P95 does not kiss with w8's affinity logic...