Gigabyte GA-H55M-UD2H OC report (update: Phase-Change)

[Massman]

 

I received this board a week ago, but only had the time today to first start testing it. First task of the day was to see how the IGP of my 661 scales in 3D. Since this particular bios (F4 - latest) doesn't offer any IGP clock frequency changes just yet, the only way to increase the GPU frequency was by increasing the BCLK. Now, that was not so easy on air cooling with the GMA HD enabled: I only got it up to 160MHz stable through 3D. Sadly enough, none of the current applications offer the possibility to see the real frequency just yet (W1zzard is working on it - due next week), so I have to settle with the easy calculation of:

 

"GPU = BCLK * Multiplier/4"

 

For the 661, the multiplier is 27x, so 133x27/4 = 900MHz. Increasing the BCLK to 160MHz, the GPU is now clocked at 160x27/4 = 1114MHz. Not bad ... but the performance still is pretty crap:

 

* 3DMark01 = 15972 - http://hwbot.org/result.do?resultId=942920

* 3DMark03 = 7116 - http://hwbot.org/result.do?resultId=942921

* 3DMark05 = 4957 - http://hwbot.org/result.do?resultId=942922

* 3DMark06 = 2594 - http://hwbot.org/result.do?resultId=943034

* 3DMark V = 772 - http://hwbot.org/result.do?resultId=943036

 

Two notes on this:

1) I am ashamed to have run 3DM01 using Windows 7 as operating system, but since no one posted in this category so far ... I just wanted a number up. I challenge anyone to beat me!

2) I changed the Intel driver setting from quality to performance and it looks like Vantage was affected by some sort of LOD. I'll try to get a screenshot of this soon.

 

Second part of the testing was to learn the board and how it can OC without the IGP, so I put in the 8800GT and tested for 2D. Max BCLK for CPU-Z was 215MHz ... 220MHz was not stable anymore ... I'm sure switching the OS will help this a bit. Because my CPU cooler isn't thát super, I had quite a lot of problems keeping the temperature within working range ... I saw 80°C regularly at which point the CPU crashed. Still, 5GHz on air was pretty darn easy!

 

 

Next up, the SuperPI 32M test to see if it's indeed that difficult to reach high memory clock frequency with high BCLK frequency. In my case, it was not that difficult, again, to do 200/2000 ... although it looks like either the memory timings are screwed up due to the bios, or CPU-Z doesn't read it correctly. I think it's a bios issue, as I have no problems when running in 2:6 mode + the performance is terrible, even for Windows 7.

 

 

Last but not least, the CPU seemed perfectly stable at 4.5G through Wprime. Fair enough for me.

 

Somewhere next week, I'll mount the single stage to this board and see how far the GMA HD can be taken :-)

[Massman]

Seems like getting blackscreen bugs on this GMA HD is also pretty darn easy:

 

[Massman]

I tested the board with phase-change yesterday and it did just fine! At first I was having issues with BCLK, but that was quickly solved by changing a couple of minor settings ... in the end, I was booting at 240MHz BCLK!

 

Temperature was around -25°C load

 

http://valid.canardpc.com/show_oc.php?id=947177

 

Memory frequency was not a problem, but sadly enough I don't have the right memory to show the high clocks. I was able to run 230MHz BCLK with 2:10 divider no problem, but was not stable enough in 32M. I'll run again once I've been able to get my hands on some decent clocking memory.

 

Some more results:

 

 

 

Still some bugs left to squash:

 

- IGP frequency scaling doesn't go that well. Even with cold, I didn't get higher stable frequency + there's some weirdness going on with manually setting the IGP frequency (is being debugged as we speak)

- Memory timings are still messed up. Either CPU-Z or the bios is doing something wrong

 

Pretty nice to see a budget board perform so well