Sapphire PI-A9RX480 Motherboard

[Tim S]

It may be Winter, but strawberries and cream are still in vogue, if Sapphire are to be believed. This motherboard, the first in Sapphire's Pure Innovation lineup, offers a fantastic implementation of the latest ATI chipset, but can it keep up with CrossFire and SLI boards? We go in-depth to find out.

[Sathy]

Interesting read, seems like a good board. Liked the colour since the first display shots were published. I've read something about sapphire abandoning the white pcb because they had some problems with the manufacturing of it, I wonder is this just gossip or?
The bundle seems rediculous, some explanation from sapphire about it wouldn't hurt!

One thing I was thinking: why use prime95 when you could use sp2004 with more options and many do consider it to be better for stress testing?
Here's the site: http://sp2004.fre3.com/

[Fly]

ouch, expensive.

[Tim S]

Originally Posted by Sathy

One thing I was thinking: why use prime95 when you could use sp2004 with more options and many do consider it to be better for stress testing?
Here's the site: http://sp2004.fre3.com/

We're not using Prime95 on its own. We're using a combination of Prime95 small FTT (working the CPU alone), while IOMeter works the hard disk controller and continuous Far Cry (or 3DMark03 depending on the reviewers preference) looping, which tests the video card and memory. These are run together, at the same time.

Boards should be able to handle it for hours on end if they are geared towards enthusiasts. However, it is worth noting that no board can handle large FTT, IOMeter and 3DMark at the same time, because Prime95 eats up 95% of your memory, when the 3D application will want the a part of that pie too.

Thanks for the link though, I will have a look at it.

[RotoSequence]

Just noting about the back panel commentary, the PCB is designed to carry an onboard X300 graphics chip with onboard RAM. The two blank spots are for DVI and VGA outputs. This board really doesnt look too terribly complete, does it.

[Bindibadgi]

Originally Posted by RotoSequence

Just noting about the back panel commentary, the PCB is designed to carry an onboard X300 graphics chip with onboard RAM. The two blank spots are for DVI and VGA outputs. This board really doesnt look too terribly complete, does it.

Yea, the PCB was designed to hold a lot of things though (not onboard ram however, no BGA solder points. Noone afaik used the LMA.)

[Reaper_Unreal]

And it was all going so well until the price. Oh well. Definatley a surprise though, didn't think that this board was going to keep up so well. If this board was sold at maybe 80, that would do it for me. That and they need to fill up all those useless empty traces on the motherboard with yummy stuff I can use, like maybe another ethernet controller.

[Sathy]

Originally Posted by bigz

We're not using Prime95 on its own. We're using a combination of Prime95 small FTT (working the CPU alone), while IOMeter works the hard disk controller and continuous Far Cry (or 3DMark03 depending on the reviewers preference) looping, which tests the video card and memory. These are run together, at the same time.

Thanks for the link though, I will have a look at it.

Yeah I've understood that from what I've read, just wanted to mention that sp2004 exist because I know for a fact that alot of tweakers and clockers use it to stress their system (and I have found it to be more versatile too) - either with other progs or just by itself.

You're welcome, it's always good to know about alternative solutions

[Bindibadgi]

Yea, tis, thanks for that But i check P95s progress by watching the LEDs on the Corsair RAM before i start FarCry. And once it's in FarCry loop there's no way to tell what's background running anyway.

[Kameleon]

Rich,

Sorry if I missed it in the article (it's kinda late) but would it be possible to add in a test for max stable HTT into the overclocking section of your motherboard tests? You know, find a CPU you know the max clock of, drop the multiplier, LDT and RAM divider, and see how far you can get the HTT. This is something people are becoming increasingly interested in especially with all these Opteron 144s that can hit 3GHz+ but only use up to 9x. I know all new S939 boards should be able to go higher than you would ever need, but it'd be nice to check.

[Firehed]

Not at all for me...

And on pg2, it says the sata ports are 150/300Mbit/sec. SATA (and other ATA specs) are MByte/sec, not bits (which on current stuff only goes to lan as far as I'm aware). Perhaps a mental switch with the 1.5/3Gbit/sec ratings which are accurate (though done differently....as of course 3Gbit != 300Mbyte, but it IS with sata thanks to the way it's done)

[Tim S]

Originally Posted by Kameleon

Rich,

Sorry if I missed it in the article (it's kinda late) but would it be possible to add in a test for max stable HTT into the overclocking section of your motherboard tests? You know, find a CPU you know the max clock of, drop the multiplier, LDT and RAM divider, and see how far you can get the HTT. This is something people are becoming increasingly interested in especially with all these Opteron 144s that can hit 3GHz+ but only use up to 9x. I know all new S939 boards should be able to go higher than you would ever need, but it'd be nice to check.

I don't know the full story but from what I understand, Rich has had a lot of issues with CPU-Z/Clockgen reporting incorrect clock speeds in recent times. The boards are not showing Rich a multiplier adjustment in BIOS for one reason or another (I don't know why), so he is unable to adjust the multiplier before adjusting the HTT/memory bus speed.

I'm planning on trying to get hold of an Opteron 144/146 for our overclocking tests, as it'd just eliminate the problems that we've been having almost completely. If there was a way to drop the multiplier to a reasonable level before getting into windows, it'd eliminate the problem completely. I just don't think he's 100% happy with Clockgen/CPU-Z.

[Bindibadgi]

The problem with the Sapphire board was that there was no clockgen for the ATI chipset. The only clockgen avaliable was for AGP versions, but ive worked out how to do stuff a little better now on AMD machines and got the ABIT board (next up for review) up to 245HTT with a 9.5 multi completely taking the CPU out of the equation. The old CG cored 4000+ is 130nm and pretty naff when it comes to overclocking so Ive tried to take it out from the equation more now, going for only HTT overclocking rather than "performance" overclocking: trying to get the most powerful balance of CPU speed and HTT. I leave HTT width and speed the same (1Ghz, 16bit up and down) because you want to test how far the board can go still.
The problem with clockgen is that sometimes the manufacturer technologies like uGuru, Corecell or other BIOS setups conflict with it and it doesnt work properly. This is why ive never been a fan of in-windows overclocking even though it is more convienient. In future I will just oc till I cant boot, then turn it down a touch and work backwards until I find a stable prime-95

Overclocking the Sapphire board was like hitting a brick wall: all would be fine then suddenly *ooof* wouldnt boot. I tried more voltage, looser ram timings, and the CPU was very little over 2400MHz so that shouldnt have been a problem.

[Nature]

Why did you use a Nvida card for an Ati board>?

[Bindibadgi]

Cause I use an Nvidia 6800GT cards for everything, have done for the last 14 months since NV were kind enough to provide us with testing samples. Unless ATI want to give me a pair of x1800s to test with, then ill use them.

[Tim S]

Originally Posted by Nature

Why did you use a Nvida card for an Ati board>?

It'd be a bit silly comparing an ATI board with an ATI card to an NVIDIA board with an NVIDIA card. We're trying to make comparisons here, and the 6800 GT was the most popular high-end card last year, by a longshot.

[Fr4nk]

Lovely Review guys, I just won this board on ebay for £90 Incp&p . So no hefty price tag for me