I.nfraR.ed

I personally use the TPU GPU clock tool, which works fine for reference cards - you can increase VMEM too. I found a passive zalman VRM heatsiink for HD4890 in my boxes, so I'm gonna use that + a strong fan. Had a great card (1130/1330 1.4625V Vgpu - '01 on water), but it died under a strange circumstances at 24/7 clocks.

How do you keep the VRM cool? I don't see heatsinks. I'm asking this, because I'm going to bench 4890 soon, but it will be my first HD4890 under cold.

It is one of the standard LittleDevil's SS. -54 on the evaporator head, I think. CPUs are from the good 3218B batch.

Yeah, mine is similar. 1.28V for 5GHz 32M, 6.93+ valid, 6.85 1M, 6.5-6.6 3D. Just for comparison - my teammate has almost the same cpu (same batch as mine) which hit 5.93 1M and 6GHz valid on SS with similar volts.

It seems that there is some bug in the XP-M 2500+ rankings - http://hwbot.org/benchmark/superpi_-_1m/rankings?start=0&hardwareTypeId=processor_1231#start=0#interval=20 I was browsing through submissions and noticed that #1 and #2 have exactly the same time and description under the score. Main screenshots are different, but the screenshot from the first submission is also present as a link in Jaan's sub. Jaan's submission screenshot shows a run at 3GHz. The competition was held long time ago, so I don't remember what was his final score, however it's listed there as a 3.3GHz run. I don't remember how was the ranking before, but I think that might be related to the recently fixed bug with unlinked screenshots to scores. Same bug might be present in other rankings as well.

I usually get my smd resistors off dead socket A boards. Most of my boards are Abit and they have many of them, you just need to know how to read values on the resistors. Congrats for your first mod, you'll certainly get it on a higher level when you get experienced enough. Just triple-measure everything before firing up the card and avoid doing mods when you're half-asleep Never measure voltage on the back of a running card, your hand can easily slip away and short something with the DMM. Don't be lazy (like I am) and solder wires for measurements. Reduces the headache, believe me

Then why these are allowed in the same stage where the lower-end Radeons aren't? Heck, even the GF4 Titaniums are not competitive enough to beat the Radeon 9700 Why Radeon/Ati guys are left with so little options, when you allow very weak cards from Nvidia at the same time? PS: thanks for the X850 cards, however already bought X850XT, but should have waited a little longer and buy XT PE instead Quite frankly, I thought the chance you'd ever respond to this is very minimal.

sandbagger

I voted for Method #2, but on a second thought Method #1 would be better. First of all, efficiency could not be taken into account, so we're left with some other option which will be a compromise. Let me think about this situation: Haswell is out and all top guys submit their 1M results, because they had it long time before launch. Then after 2-3 weeks I'm able to purchase my retail cpu when it gets to Bulgaria. I push to the max with the best efficiency I can achieve and eventually match some of the inefficient scores, but get lower points, because I was not able to submit earlier. How's that exactly fair? Efficiency can not be taken into account, we all agree with what Pieter said on the topic. Now, averaging the points for the tied results will eventually make the other guy who submitted first to push harder and earn his points again. And that's a good thing. The thing is older scores (e.g. Athlon64 3200+ 1M rankings) submitted like 5 years ago might not be able to "defend" their position, but I still think it's the better option than the current one. BUT, if there's a significant performance hit for the server, then I'd leave it like it is now. Looking at the 3770K 1M rankings, there are too many ties between 5-6 guys at every step (5.234, 5.250, 5.281 and so on). Maybe it's a good idea to have that average points in case of a tie for top5 only in popular rankings, because the point difference there is bigger. For other places it's like 0.5pts max.

It drops when you have CCC and/or MOM processes running.

Wrote him on msn, let's hope we'll get answers soon. Meanwhile, if someone has more free time maybe it's a good idea to summarize all unanswered questions in a single post. Othewrise I'll do it when I get back home. 850Pro/XT/XT PE PCI-E | SC2 #3 9100(8500LE)/9000/9000Pro | SC2 #2 QX6700 | SC2 #4 Prescott-256 (Celeron D) | SC2 #3

Pieter, I know TeamCup is out of scope, since you are busy with the ProOC and it gets all the attention, but please, can we ordinary overclockers get answers for the questions which were posted over and over again on the previous pages?

Mine is running -120 or so, then the memory bugs out, I think.

Nice card. I think your core clocks better than mine

Vapor-X version is different than the reference design and needs volt-mod. It uses analogue vrm.

Still no clear decision about 850Pro/XT/XT PE PCI-E and 9100(8500LE)/9000/9000Pro

People are just clicking on the "Add your benchmark" link at the bottom of the page, thinking this is the way of submitting a score.

Any thoughts on this? IMO it black-screened in GT2.

Where do you get your info for X850XT/XTPE/Pro/ from? I see various info on different sites. Article from Tom's hardware (2004) - http://www.tomshardware.com/news/ati-launches-x850,413.html Wikipedia (2005) - https://en.wikipedia.org/wiki/Comparison_of_AMD_graphics_processing_units#Radeon_R400_series TechReport tested both X850XT and XTPE on their review from Dec 1 2004 http://techreport.com/review/7679/ati-radeon-x850-xt-graphics-cards Not sure which of them were available in stores at the end of 2004 if any of them. And even that is not a criterion, because we all know availability in different markets is different. IMO they should be threated equally (Pro, XT and XTPE PCI-E versions) - either allow or disallow.

own bios mod based on official D27.

This is my best Tbred out of 15+ different models. The goal was to beat IceAngel's LN2 result - a little challenge for myself.

According to my testing there was no difference in the score, but the newest CPU I've tested was Thuban. There might be a difference with Bulldozer/Piledriver.

Nice, but I'm sure you can do better or have backups

GT2 is too high. Over 130 fps more compared to other scores in the rankings.

The cpu just doesn't go much further than that, so I took the other route with high fsb, but it's hard with Thunderbird.