Bulldozer on Asus Crosshair IV Extreme (BIOS 3027)

My initial testing with BIOS 3027 on the CHIVE. The following does not work:

Per core multiplier settings in TurboV (changing multis in turbov had no effect on CPU frequency)
Disabling cores/modules (no options, I could set CPU core activation to manual or auto, but manual offered no settings at all)
Disabling power-savings (with all power saving features turned off, the CPU still downclocked at idle, making it impossible to do a real CPUz run as I would need to load the CPU with superpi to get a frequency reading)
Manual Voltage Settings for CPU and CPU-NB (only offset voltage settings were available)

So, BIOS 3027 brings bulldozer "support" to the CHIVE, however it is fairly basic - it would probably be fine for a daily driver, but it makes the board totally worthless for benchmarking or subzero overclocking.

That said, it did run my FX-6100 at 7.080GHz (1.824V) and complete superpi1m almost automatically, I didn't have to tweak at all. I did not feel it was having any problem delivering enough power to the chip, and if BIOS support was better, I could have tried to push it. With the problems above though, there's no point to pushing Bulldozer on this board.

Maybe I will hold out a little hope that the full support that was supposed to be, will eventually become full support

I've been running BD on my CHIVE since BD came out (well, 3 or 4 days after it came out), and have it stable at 4.7886. I'd like to get it over 5 GHz, but cooling is driving me up a wall. Short of going to something more exotic then the simple Antec Kuler H20 920 I have been using, I don't see anyway to keep stable running Prime95 any higher then I am now.

I'm using 5 fans on full in the Lanboy Air case, so ambient temps in the case are about as low as the house temps, except exactly where the CPU mounts on the MB. Can't put a fan there, not enough room without putting it outside the case -- Which I don't want to do.

Oh, about not keeping clock locked... Have you not disabled "Cool & Quiet"? Seems to work on my setup. CPUz stays on 4768 MHz when C&Q is disabled.

-Rodger

you are doing pretty well on that chip from what ive seen.

I had disabled cool and quiet, as well as all the other options on the CPU settings BIOS page. not sure why the settings didn't stick, what BIOS version are you on?

I.M.O.G. said:

you are doing pretty well on that chip from what ive seen.

I had disabled cool and quiet, as well as all the other options on the CPU settings BIOS page. not sure why the settings didn't stick, what BIOS version are you on?

Click to expand...

Well, running Prime95 as I type this and see I have errored. CPUz is indeed showing CPU clock cycles fluctuating. Guess I wasn't paying too close attention before -- Please forgive.

I'm using the 3027, dated 10/28/2011. I see there is another 3027 dated 11/18/2011 however. I didn't think they re-released BIOS updates without incrementing the number, but maybe they did in this case. Probably worth flashing just to see if anything is different (bettor or worse).

-Rodger

I did a file compare on the 2 versions of CHIVE BIOS ver. 3027 I have and both files are identical. I don't understand why they have 2 different original (creation) dates. Anyways, no differences, so no reason to flash.

-Rodger

The Problem

Made a new discovery last night concerning my Antec Kuhler H20 920. I had been just too mystified as to why there was such a difference between the CPU hardware temps and the coolant temps under prolonged load. I've been seeing right around a 40 C. differential, which just seems too much to me. I'll admit I'm new to water cooling, as the potential cost had been more then I was willing to spend, until these self-contained units became available a while back. But that said, I'm not new to thermal overload on CPU's/components breaking things. Seeing what my Noctua NH-U12P did for temps on my Phenom II x6, I figured most likely it wouldn't be enough of a cooler for this BD upgrade, so in came the 920. In my sometimes child-like mind I guess I was expecting to push 5 GHz at stock temps I suppose, but I sure wasn't expecting hitting 80 C. socket temps with water!

Tried to iron this out by calling Antec tech support, but they haven't been trained (their words, not mine) on water cooling yet -- so no help there. I sent an email to the manufacture of this cooler, but they referred me back to Antec, so no help there either. Sort of left this up to me and my own sometimes warped thinking... Which sort of worked out in this case, I think.

Been giving you readers clues here, have you caught on yet?

Well, figuring something just ain't right, I decided to take the cooling block off and have a look at how well the TIM was dispersed across the CPU. I knew I had put the Arctic Silver 5 on properly, but just had to see for myself to make sure -- Well, was I surprised at what I saw.

Looking at the top of the now exposed CPU, I could see a decently spread out, thin layer of TIM across the entire CPU, well nearly that is. There was a circular spot directly in the center of the CPU, about 1/2 inch in diameter that had only a micro thickness of TIM. So little in fact, it was all but transparent!. Realizing I had not applied the TIM in any manner to allow for this, it was obvious at that point some external forces had been at work here.

Upon looking at the cooling block mating surface, sure enough that same effect was apparent. I could clearly see the spirals machined into the shiny copper face, right though TIM. Only one conclusion came to mind at this point -- Obvious to me, this cooler block wasn't flat, but had a machined crown right in the center. Now I don't know everything, so I guess I didn't know if this is suppose to be or not. But upon cleaning off all the TIM from the CPU and cooling block, I was able to hold the edge of my dial indicator on both the top CPU and the shiny copper spiraled bottom of the cooler block. Sure enough that nice shiny copper surface was indeed machined with a nice, uniform crown at the center, about .50mm up from the edges. The CPU surfaces was nice and flat.

Is this right?

That's the question I was asking myself for a few minutes. As I had said, I'm new to water cooling, so figuring I don't know all the "tricks" yet, I had to think of all the possible "whys" they would have purposely machined in this nice, shinny crowned, not flat mating surface on this cooling block. For the life of me, nothing was coming to mind here, except a 2 single possibilities.

1. They obviously made a mistake here.

2. They obviously think most people will use to much TIM and this is the method they thought of to disperse the excess TIM away from that 1/2 inch thermal sweet spot.

Since both of these "obvious" conclusions could not be true, I immediately ascertain what must be done -- FIX IT! My Tim Taylor persona was instantly activated and so was my "not so warped" thinking.

See part 2 -- The fix. Will post later after work.

That is the way it should be actually, you don't want hardly any TIM between the mating surfaces, and good contact near the center of the die is the most important. The base of some coolers are normally very slightly convex, and that is ok - if it were concave, that would be a real problem.

Some people lap things flat, and as long as both the CPU and heatsink are good and flat that is ok too. But most people consider a slightly convex base not such a bad thing I believe.

Well, big letdown folks. Had this cooler about 4 C. cooler at peak load. Figured it was worth the effort of fixing the cooler block, but alas the pump died today. What a deal.

What I did was exactly what IMOG said, I lapped the matting surface of the cooler block to within a nats whisker of flat. Buffed it up to a nice shine and put it back on the CPU. The most dramatic thing was how much faster temps drop after stopping a heavy load. I'd have to say about twice as fast as it did before the fix.

Now I'm torn between replacing this one with the same thing, or trying the Corsair brand.

What would you do?

-Rodger

Aren't they both made by the same original manufacturer, and rebadged? I forget as there are a number of these sort of coolers out now, but several of them at least are made by the same company. I'd look into who the original manufacturer of the components is first, then I'd choose by price, and consider features last - different brands may have a few different features or software options.

I.M.O.G. said:

Aren't they both made by the same original manufacturer, and rebadged? I forget as there are a number of these sort of coolers out now, but several of them at least are made by the same company. I'd look into who the original manufacturer of the components is first, then I'd choose by price, and consider features last - different brands may have a few different features or software options.

Click to expand...

Yes, I believe both some of the Corsair products are and Antec's are from the same manufactue -- Asetek. They'er the ones I sent an email to about the thermal transfer not being as good as I thought it should.

Went to Bestbuy and purchased a Corsair H60 water cooler. Figured I try it and see how well it compared to Antec's Kuler 920. I was pretty surprised to find it does quite well. I had to drop my buss to 260 from 265 to keep stable, but temps are doing very close to my last prime95 run using the 920.

Apparently (for all I know anyways), water cooling will not drop CPU temps very much more then air does, just hold temps low longer. Too bad there isn't a cheaper way to use chilled water, eh.

-Rodger

marjamar said:

Apparently (for all I know anyways), water cooling will not drop CPU temps very much more then air does, just hold temps low longer. Too bad there isn't a cheaper way to use chilled water, eh.

-Rodger

Click to expand...

Think of it this way Rodger... Water cooling is just a way to "air cool" your rig, but instead of having a size/weight limitation where you can't just keep on getting bigger and bigger heatsinks to fit on top of the CPU; you just move the bulky heatsink part away from the problem area (in the form of a radiator) and use water as the transport medium used to move the heat from the CPU to your big heatsink (radiator) somewhere else.

Water won't bring temps down further than air, at least not anything noticeable, but they will cool higher loads than air coolers can... This is due in part to the larger size - surface area that radiators have to dissipate that heat load.

There are "cheaper" ways to do chilled water. You just have to get inventive. Look for water chillers at aquarium stores, bring the radiator outside in cold nights, etc. Just remember that with chilled water you will have to address the whole condensation issue.

xXSebaSXx said:

Think of it this way Rodger... Water cooling is just a way to "air cool" your rig, but instead of having a size/weight limitation where you can't just keep on getting bigger and bigger heatsinks to fit on top of the CPU; you just move the bulky heatsink part away from the problem area (in the form of a radiator) and use water as the transport medium used to move the heat from the CPU to your big heatsink (radiator) somewhere else.

Water won't bring temps down further than air, at least not anything noticeable, but they will cool higher loads than air coolers can... This is due in part to the larger size - surface area that radiators have to dissipate that heat load.

There are "cheaper" ways to do chilled water. You just have to get inventive. Look for water chillers at aquarium stores, bring the radiator outside in cold nights, etc. Just remember that with chilled water you will have to address the whole condensation issue.

Click to expand...

I looked at the aquarium chillers last year when I first built this video editing system. I had it running on air cooling using the Noctua NH-U12P and didn't like being right on the boarder temp-wise, so was thinking water cooling. Seeing how much it was going to cost for water ($500 to $750) for full blown system, I started thinking about perhaps chilling the water down too. Got too frustrated thinking about all the maintenance water brought with it, then the added effort to keep condensation levels managed if I went with chilled water, and more or less just set all of it aside.

When I did the BD upgrade to this system, I knew the Noctua would have more problems on this chip then the Phenom II 1090T, so I thought I'd try a closed water cooler, hence the Antec Kuhler H20 920. Since that broke I'm now on Corsair H60 which is a bit less capacity-wise, so I had to back the bus down from 265 MHz to 260 Mhz (still at x 18 multiplier).

Right now I'm fully stable (Prime95 will run as long as I let it) @ 4.7 Ghz. Got memory working as good as it will with this processor, so if Asus would just update BIOS to fix a couple of things wrong, I think this computer upgrade is finished. If they don't fix the BIOS (or wait too long to fix it), I may have to moving into another MB, then of course I need to start the "Tweaking" all over again.

-Rodger

So been doing a bit of testing with this little H60 water cooler block. It's really not as bad as I was expecting it to be, but it's capacity is too low for a sustained Prime95 torture test running this BD 8150 @ 4.7GHz. The test will run as long as I let it without errors, but the CHIVE's VRM kicks in and drops Vcore and multiplier to cool it down. This processor will run stable at alot higher temps then the 1090T would.

I've got a Corsair H80 coming in, so I hope this will keep the CPU cool enough to maybe limit this from happening. I might need to look at getting some more, cooler air over the VRM fan.

Look at the graph on the left of the pic to see this drop in temps. Just before the drop occurs, the Vcore drops and a second or 2 later the multiplier drops. I does this in 2 stages, first sets the CPU to 4.1 GHz, the next sets it to 1.8GHz a second or 2 later.

Aside from the fact this is hard on the CPU, it is none the less impressive on how well it actually does this without causing any hard errors running Prime95.

-Rodger

BTW - I use Goolge Chrome and to see these large pics easier (can actually read the text), I right click on them and choose "Open image in new tab".

I edited your post and used the paperclip icon to insert the image inline, rather than attached to the post. If you used the paperclip icon with your images, the forum software automatically resizes big images.

Interested to see how the H80 fairs.

Awso.

Got the H80 installed last week (after waiting about 8 days on a shipping mix-up) and seemed to work pretty well... Up to a point. Running Prime95 at 4.7GHz, it would get too hot after just a few minutes.

I had already posted this in another thread, so rather then go over the same ground twice, CLICK HERE to go to that thread. Click the top-right corner (Thread: [FRONTPAGE] AMD FX-8150 - Bulldozer - Processor Review) to continue reading the remainder of the post.

Sorry for making anyone click to read the rest of this story on my BD install on CHIVE.

Rodger