Epox 9NDA3+ Socket 939 NF3 Ultra board

[Omega65]

Originally posted by Bluetooth:
The increased FSB won't bring your anything more because the multiplier has to be adjusted downwards.

Actually 8x 250mhz = 2.0ghz should outperform 10x 200mhz = 2.0ghz.

But I was suggesting Max FSB/Max Clockspeed benchmarks.

[pweltz]

Looks like a nice board; the primary sata ports are in an unfortunate position though....almost sure to interfere with Nvidia 6800 Ultras or ATI cards with Artic Cooling-style dual slot heat sinks. The wide chipset cooler is nice....every Nforce3 board I've seen has the chipset so close to the AGP slot that installing an aftermarket cooler is pretty much impossible.

[AndyB]

Very nice review! And what a fantastic mainboard...
Two small comments:
1) Hardware RAID is not so important for me, because I use Linux and Linux has a very efficient software RAID implementation. I can even have RAID across SATA and PATA ports.
2) I see Epox has finally noticed that most people don't use the PCI slot next to the AGP slot, and used that space for the lithium battery. I wish more mainboard manufacturers would follow the example.

[fredyama]

Thank you Bluetooth for your excellent review. [img]smile.gif[/img]
Looking forward to get this MB to make some performance comparison with 8KDA3+. :D :D

[bluetooth]

8KDA3+ is a socket 754 board whereas 9NDA3+ is a Socket 939 board. you can't compare it this way.

[Omega65]

Originally posted by Bluetooth:
8KDA3+ is a socket 754 board whereas 9NDA3+ is a Socket 939 board. you can't compare it this way.

But A comparison I would like to see is a Athlon 64s benched vs each other.

1) A64 3400+ 2.4ghz 512K S754 vs A64 3800+ 2.4ghz 512K S939
2) A64 FX-53 2.4ghz 1MB S939 vs A64 3700+ 2.4ghz 1MB S754
3)A64 3400+ 2.2ghz 1MB S754 vs A64 3400+ 2.4ghz 512K S754 vs A64 3500+ 2.2ghz 512K S939

It would be nice to see what matters more - Single/Dual channel, Clockspeed or Cache.

IMHO in order of importance (to performance) It's 1st Clockspeed, 2nd Cache, then Socket type last

[RayvenX]

It varies too much to test fairly... The extra memory bandwidth of 939 CPUs in some cases could make a huge difference, yet for other apps will make nearly none, and the same with more cache vs. less.

Anyway, I don't think BT has a working 754 CPU anymore

[bluetooth]

If you can propvide me with all the xpensive CPUs, I can do a test

[RayvenX]

Originally posted by AndyB:
Very nice review! And what a fantastic mainboard...
Two small comments:
1) Hardware RAID is not so important for me, because I use Linux and Linux has a very efficient software RAID implementation. I can even have RAID across SATA and PATA ports.

That's not entirely a good idea: You can't boot directly from RAID 0 software partitions under Linux, and CPU utilisation for anything more complex than RAID 0 or 1, or 0+1 is uslessly high for most purposes (Such as RAID 5, where RAID actually starts to offer a useful performance benefit AND a useful integrity benefit)... Hardware RAID still has a long life ahead of it.

2) I see Epox has finally noticed that most people don't use the PCI slot next to the AGP slot, and used that space for the lithium battery. I wish more mainboard manufacturers would follow the example.

They're hardly the first people to realise. My DFI doesn't have the slot below the AGP one present, in fact it has the AGP slot in the highest possible position, unlike most boards and thus still has 5 PCI slots.

[Omega65]

Originally posted by RayvenX:
It varies too much to test fairly... The extra memory bandwidth of 939 CPUs in some cases could make a huge difference, yet for other apps will make nearly none, and the same with more cache vs. less.

That's exactly what would be nice to know...does it Really vary all that much. Especially considering the price differntial between (say) the S754 3400+ and the S939 3800+ (Same clockspeed and cache). If the difference averages under 5-6% then the 3800+ may be overratted (or the 3400+ Newcastle underrated)