uATX Results

[JKing76]

Why would you want an integrated-video chipset if you're going to immediately put in a dedicated graphics card?

 

[Gary Key]

Update, we have encountered additional technical difficulties today, Jarred and I are busily editing the article now and continuing to code the charts. The article will be up as soon as we are finished, it might be midnight, might be 4am... might be first thing in the morning... more later..

 

[Zap]

Originally posted by: LTE
Between the DS2R and S2, the same thread notes that Raid, # of sata ports, and solid capacitors are the only differences. What difference does solid capacitors make?

Longetivity. IMO the most problematic area for capacitors are the CPU power, which is solid on all these boards. I'd just get the cheaper board as long as you don't need RAID or the extra 2 SATA (assuming same BIOS).

 

[licuo]

I see an article posted on the front page, but the link is invalid!

 

[Gary Key]

Originally posted by: licuo
I see an article posted on the front page, but the link is invalid!

One of the technical difficulties we had yesterday, it is fixed and the charts are going in now, the article will go up this afternoon. MCP73 preview is early next week, another P35 board, and then on Thursday now, the AMD 690G/NV7050 boards go up, then a follow up the next week with another G33 board, two G31s, and a couple of revised G965 boards. After that we will start our audio, video, chassis, and Linux articles centered on uATX.

 

[LTE]

Originally posted by: JKing76
Why would you want an integrated-video chipset if you're going to immediately put in a dedicated graphics card?

It's more the form factor, to be honest, and microATX by and large seems to come with integrated video. Maybe I'm completely off about that; but even if I am, the G33 seems to be an excellent chipset, so no big loss.

 

[Zap]

Woot! It's up! Too bad I have to step out for a while... will read it later this evening.

 

[Skott]

I was looking to see the DS2R slightly edge out the P5K-VM in overclocking since thats what many people are experiencing it seems. A surprise but a pleasant one. Just makes me feel warm and fuzzy I made a good choice when i bought my P5K-VM.

 

[hans007]

the g33 and p35 are the same chip. die size is the same, and TDP is the same.

thats why g33 boards overclock well, the p35 is just a crippled g33.

 

[SerpentRoyal]

None of those G33 boards were able to extract much speed from the 2160 CPU (800MHz chip). Could be a bum chip.

 

[gorobei]

Originally posted by: SerpentRoyal
None of those G33 boards were able to extract much speed from the 2160 CPU (800MHz chip). Could be a bum chip.

the strap modded 2160 did better. [Any thoughts on how the 4x00 chips might do? or will all 3.0+ overclocking require a pin mod?]

The gigabyte bios will likely go through a couple more versions before we see any significant increase in OC, if the 965 boards are any indication.

I'm probably going to wait for the G35 boards to come out. But that asus p5k-vm is very tempting.


(PS: I AM planning on a discrete graphics card and a large tower HS. The IGP, if it does HD output, is just gravy. Otherwise its just a backup if the videocard fails.)

 

[chucky2]

Hey Gary, just wondering, when you post the AMD 690G stuff, will any of the boards be running the "performance BIOS" that AMD released to the manufacturers?

If not, is there any reason why it's not out yet?

Thanks either way!

Chuck

 

[hans007]

Originally posted by: JKing76
Why would you want an integrated-video chipset if you're going to immediately put in a dedicated graphics card?

i alwyas get integrated video (i also always get microatx since i like the small cases).

reason being, integrated video is more or less half decent now an x3000 is maybe as fast as a 7100gs geforce.


i am not much of a heavy gamer, so i generally just buy a decent card when a game i want to play comes out (since i probably play like 2 games a year) and then sell the card when i get bored of the game.


it is also nice to have the igp there, so that when you eventually upgrade the board and say want to use it in a box that doesnt need any 3d you dont need to buy a video card again.

 

[SerpentRoyal]

Originally posted by: gorobei

Originally posted by: SerpentRoyal
None of those G33 boards were able to extract much speed from the 2160 CPU (800MHz chip). Could be a bum chip.

the strap modded 2160 did better. [Any thoughts on how the 4x00 chips might do? or will all 3.0+ overclocking require a pin mod?]

The gigabyte bios will likely go through a couple more versions before we see any significant increase in OC, if the 965 boards are any indication.

I'm probably going to wait for the G35 boards to come out. But that asus p5k-vm is very tempting.


(PS: I AM planning on a discrete graphics card and a large tower HS. The IGP, if it does HD output, is just gravy. Otherwise its just a backup if the videocard fails.)

A good P35 board has no problem overclocking these 800MHz chips up to 390MHz FSB range using 9x or 8x multi. Clearly this is an issue with G33 boards.

 

[Zap]

Originally posted by: gorobei

Originally posted by: SerpentRoyal
None of those G33 boards were able to extract much speed from the 2160 CPU (800MHz chip). Could be a bum chip.

the strap modded 2160 did better. [Any thoughts on how the 4x00 chips might do? or will all 3.0+ overclocking require a pin mod?]

Strange, I did not see results on the BSEL modded E2160 for the MSI board. Oversight? The Asus and Gigabyte boards both gained 65MHz FSB, but nothing reported for the MSI. If it has similar gains, then the cheap board is "teh winnar" for me.

I'm thinking this has to do with FSB strap.

I've also encountered issues overclocking an 800MHz FSB CPU (E4300) on certain boards due to not having a lower memory multiplier, and thus causing the memory to be pushed too far.

Originally posted by: hans007
reason being, integrated video is more or less half decent now an x3000 is maybe as fast as a 7100gs geforce.

For 3d? I don't think it is quite there yet. It may do fine in some things (like Aero and 3DMock) but performance on Intel IGPs is erratic. The 7100GS "should" have more consistent performance and better IQ. Of course Gary didn't test IGP versus low end GPUs so we don't know that (from his article).

Originally posted by: SerpentRoyal
A good P35 board has no problem overclocking these 800MHz chips up to 390MHz FSB range using 9x or 8x multi. Clearly this is an issue with G33 boards.

It is probably a BIOS issue since the BIOS has control over FSB strap, internal timings and memory multipliers to the extent of what is available in the chipset. Just the fact that a bit of conductive ink on the CPU can suddenly push the overclock from the 2.8GHz range into the 3.4GHz range is pretty telling.

That's one thing I've seen in Abit boards in the past... FSB strap option in the BIOS. Don't know if current boards have it (SerpentRoyal?) but it had helped me with my Abit IS7 way BITD when overclocking low FSB chips.

 

[SerpentRoyal]

IP35 boards don't have separate strap adjustment for overclocking. 1:1 and 1:1.20 dividers are on 1333, while 1:1.25 and 1:1.50 are on 1066. Note that all IP35 boards with stock BIOS can easily hit 425MHz FSB with a capable 800MHz chips. These boards have no problem overclocking 800, 1066, or 1333 chips to 98% their maximum potential. I didn't encounter any FSB hole with my E4300 and E6320 up to 488MHz FSB (limit of E6320).

I'm not convinced that this is a BIOS issue. All three boards from three manufacturers had difficulty with 800MHz chips. Perhaps it's because they all share the same source code.

The inability to take these VALUE CHIPS to their full potential is very disturbing. All C2Ds are on par at the same core speed. Why pay $150 for E6xx0 when you can overclock a $60 E21x0 chip to 3.2GHz?

 

[Skott]

What AT got in OCing isnt indicative to what a user will get necessarily. I know people have gotten better overcloks on the DS2R and the P5K-VM than what Gary got. The HF has alot more info about these two mobos than anything AT has put out. AT has been very late with this info and late to the party. The main thing they did differently is pit all three together in one review/comparison. Hopefully we'll see others do this as well to get a broader picture.

Edit: I should mention the XS forum has had more info out longer than AT too.

 

[hans007]

Originally posted by: Zap

Originally posted by: gorobei

Originally posted by: SerpentRoyal
None of those G33 boards were able to extract much speed from the 2160 CPU (800MHz chip). Could be a bum chip.

the strap modded 2160 did better. [Any thoughts on how the 4x00 chips might do? or will all 3.0+ overclocking require a pin mod?]

Strange, I did not see results on the BSEL modded E2160 for the MSI board. Oversight? The Asus and Gigabyte boards both gained 65MHz FSB, but nothing reported for the MSI. If it has similar gains, then the cheap board is "teh winnar" for me.

I'm thinking this has to do with FSB strap.

I've also encountered issues overclocking an 800MHz FSB CPU (E4300) on certain boards due to not having a lower memory multiplier, and thus causing the memory to be pushed too far.

Originally posted by: hans007
reason being, integrated video is more or less half decent now an x3000 is maybe as fast as a 7100gs geforce.

For 3d? I don't think it is quite there yet. It may do fine in some things (like Aero and 3DMock) but performance on Intel IGPs is erratic. The 7100GS "should" have more consistent performance and better IQ. Of course Gary didn't test IGP versus low end GPUs so we don't know that (from his article).

Originally posted by: SerpentRoyal
A good P35 board has no problem overclocking these 800MHz chips up to 390MHz FSB range using 9x or 8x multi. Clearly this is an issue with G33 boards.

It is probably a BIOS issue since the BIOS has control over FSB strap, internal timings and memory multipliers to the extent of what is available in the chipset. Just the fact that a bit of conductive ink on the CPU can suddenly push the overclock from the 2.8GHz range into the 3.4GHz range is pretty telling.

That's one thing I've seen in Abit boards in the past... FSB strap option in the BIOS. Don't know if current boards have it (SerpentRoyal?) but it had helped me with my Abit IS7 way BITD when overclocking low FSB chips.

supposedly the x3000 with the newerst drivers in XP is pretty close to a 6200tc/7100gs (same card). most of the charts i've seen put it within 5%

 

[Zap]

Originally posted by: hans007
supposedly the x3000 with the newerst drivers in XP is pretty close to a 6200tc/7100gs (same card). most of the charts i've seen put it within 5%

Then the next question is can it perform that well while retaining image quality? On some reviews comparing various IGPs, the Intel often was the slowest as well as lowest image quality.