I.M.O.G.
-
Posts
869 -
Joined
-
Last visited
Content Type
Profiles
Forums
Events
Blogs
Posts posted by I.M.O.G.
-
-
Quit screwing around... Grab a sandwich bag and dry ice and get that sucker subzero.
We did an article on that a couple years ago:
http://www.overclockers.com/extreme-gpu-cooling-extreme-budget
-
How about "freezing" the points fo benchmarks that are removed? So when a benchmark is "not active", the global points for the submissions that were submitted will be frozen and not lost. You can still submit results for non-active benchmarks, but you will not get points for it.
This way, voting a benchmark out will not make you lose the points you worked for in the last couple of years.
Not this. "Safe" or "unbeatable" points don't exist currently. If a bench isn't active, those would be the most valuable rankings to have because those are the only points that don't decay as others beat you.
That would not give the intended affect I don't think. Pcm05 is almost like this currently. Very low competition, easy points that almost don't go away when you get them because no one but a few want to run it.
-
Impressive.
how do you get the validation links for this? amazing clock speeds. nicely done. congratsYou save a validation file with CPUz, then you upload it here to get a validation link: http://valid.canardpc.com/
-
This really doesn't matter, so I never mentioned it, but I notice it EVERY TIME I view the rankings and its annoying. heh
When viewing a country ranking, like this one:
http://hwbot.org/league/oc?offset=-7&countryCode=US
The H3 tag says "The Fastest Extreme Overclocker of the World", even though I'm looking at the US XOC rankings. It could use the CountryCode from the URL to say "The Fastest Extreme Overclockers in the *CountryCode*".
It really doesn't matter. But its bothered me for as long as I remember.
-
-
Cool, looking forward to your next subs. You full pot on your card?
-
Thanks guys. This is the first GPU I ever benched, and still my favorite one. (Same one bobnova repaired a busted SMC on)
Amateurs - let's see those backups. :-D If you top this, I will be rerunning - seemed as though the card had more, this is just where I stopped.
Unfortunately AM3 was the only bench I got in last night.
-
Hardmods necessary to go beyond 1.5? That's where afterburner stops for me.
Your score made me rerun tonight. I couldn't hit your CPU clocks, but my GPU clocks were much stronger than last time.
Keep pushing! These cards are fun.
-
Great score.
Probably a reference card l_s, the clocks are similar to what I've ran on my 4890 which is all stock - I ran it with higher GPU and lower mem.
-
No idea
I didn't check the breakdown, I got the kinda score I was expecting, so I saved it. Not many guys have benched 4850/70/90 with Ivy. Give me a sec, i'll check the leaderboardEDIT: Checked. My score "pattern" doesn't match any of the top 7 scores, but there are two, COMPLETELY different kinds of CPU and GPU score mix, so i'm not especially worried
EDIT 2: my chapter scores are fairly normal, but the CPU and GPU scores look TOTALLY random
Your score looks right to me, despite it being displayed in the wrapper funny . I ran 4890 on a 6.2GHz Ivy and its pretty close to your score, except I had a bit extra CPU and GPU clocks.
http://hwbot.org/submission/2294253_i.m.o.g._aquamark_radeon_hd_4890_443866_marks
-
You still going to Canton?
They are paying a few cents more from distribution since last month in Bedford but still only charge me $.50/L.
Yes, still canton. Bedford takes forever to fill my 150L. Might start going back though. Gonna try Youngstown next I think.
-
1 Euro in our money is 40,000Rials
1 Dollar in our money is 30,000Rials
LN2 price is 7,000 Rials per Litre In Tehran
we can take shower with Ln2
Nice, thats really good.
The price of generating LN2 is basically whatever the electricity costs to run the compressors - I'm guessing Tehran has a pretty good price per KwH.
LN2 prices are a bunch of crap in the first place, the margin is ridiculous unless you are really far from the generation plant. Being a byproduct of the production of Liquid Oxygen, which is used in huge quantities in industrial and medical uses, they might as well bottle up the resulting LN2 and sell that too.
So basically, the cost of generating LN2 is actually less than the price it costs for the electricity to run the compressors, because the cost of electrity is divided among the other liquid gasses generated in the process.
So if they give you some crap about it being expensive stuff, explain to them how its made, and what their cost is to produce it. Their only real cost is keeping the machines running, paying someone to deliver it, and paying someone to take your money when you take it off their hands.
-
I got a cost increase. Previously I posted about $.50/L. They bumped me to $.55/L and I did my best negotiation to ensure they didn't take me to $.60/L though they tried. I usually buy 150-200L at a time, and when I have enough time, I buy 400/L a month. Fortunately my prices aren't as bad as many people have, but I've shopped around and done a lot of negotiating... But a 10% price increase is a lot, and I'm really glad they didn't get me for the 20% they were shooting for. It would have cost me about $40 extra a month for the same amount of stuff, plus every trip in the truck to pick it up costs me another $15 round trip in gas, along with 2 hours drive/fill time.
Prices are for my own dewars getting filled (150L + 50L), and my own transportation to/from facility.
-
Graphs look great. The title above the graphs is listed as "Team Points History", regardless of which graph is being displayed (team points, team members, achievement, team rank)... Maybe a more general heading there would make more sense, or update the title as different graphs are displayed.
-
Nice bigrig subs bud. You just forgot to click on the memory tab on this one, but I see you got it right on wprime1024.
-
This one is no good either (old cpuz version, no validation link): http://hwbot.org/submission/2341619_alfagfx_cpu_frequency_core_i5_2500k_6009.9_mhz
EDIT: Actually, everything in his profile is submitted wrong, or faked.
-
Well, the upcoming 3DMark will probably then be interesting in this regard...
Makes more sense to integrate frequency monitoring into the benchmarks.
Agree that it would be a good thing especially in 3D where static frequency isn't very realistic in the top rankings of any hardware category.
-
How do you insulate for cold?
What do you put your motherboard on when benching?
I bought an MVG originally, benched without cold bugs for a bit, then it started coldbugging constantly and I couldn't get them to go away. So I bought a second MVG, benched without cold bugs for a bit, then it started coldbugging just like the first one. There were a couple theories on why:
- I was benching them on a motherboard box, which was causing some flex on the motherboard PCB, and damaging the socket area
- I was insulating with petroleum jelly, and it was contaminating the socket, causing contact/conductivity problems under cold
A friend of mine also had an MVG and benched with me at my place, and it also developed cold issues under LN2. Another member on my team had the same issues on an MVG. I have heard of others not having these problems on the MVG - it is no guarantee every MVG will have this problem, but it is a common issue.
I moved to the Maximus V Extreme, and I have had fewer problems. I still had cold bug issues on occasion, but could usually work past them with continued effort. 2 weeks ago, I had cold bug issues on the MVE that I couldn't get around. So I melted off all the petroleum jelly on the board, cleaned out the socket with a hair dryer and a paper towel, and the next time I ran it I had no cold bug issues at all.
So now I've since stopped using vaseline for insulation, and just use a little frost king, paper towels, and a lot of high CFM fans to keep the board warm. I've also stopped benching on motherboard boxes, and I elevate the board with a 1/2" of armaflex foam so the entire board is firmly supported.
Dunno if this helps. Personally, I really like the Asus bios, and I have less trouble getting clocks up on Asus boards in my experience. If I were you, I'd probably look at an MVE. If that is price prohibitive, you might try another MVG, but I've just had and heard enough bad experiences on cold with them that I couldn't recommend it really.
FWIW, I sold my MVG to someone who just runs air, and its been running great for them ever since. I told them I had problems on cold.
-
When searching submissions, Piledriver CPUs are shown as Bulldozer. When comparing Piledriver vs Bulldozer, it is better if one can search separately by subfamily. The piledriver subfamily doesn't exist, and currently piledriver CPUs are classified as bulldozers.
[noparse]http://hwbot.org/search/submissions/permalink?applicationId=3&scoreOperator=more_or_equal&score=&username=®ionCode=&countryId=&teamId=&glPointsOperator=more_or_equal&glPoints=&hwPointsOperator=more_or_equal&hwPoints=&compPointsOperator=more_or_equal&compPoints=&globalTeamPowerPointsOperator=more_or_equal&globalTeamPowerPoints=&hardwareTeamPowerPointsOperator=more_or_equal&hardwareTeamPowerPoints=&totalPointsOperator=more_or_equal&totalPoints=&league=&globalRankOperator=less_or_equal&globalRank=&hardwareRankOperator=less_or_equal&hardwareRank=&cpu=&cpuId=&[/noparse]cpuSubFamily=Bulldozer[noparse]&cpuSubFamilyId=204&cpuFamily=&cpuFamilyId=&cpuBatch=&cpuBatchId=&cpuFreqOperator=less_or_equal&cpuFreq=&numberOfProcessors=&cpuSocketId=&cpuCoolingId=&gpu=&gpuId=&gpuFamily=&gpuFamilyId=&gpuBatch=&gpuBatchId=&gpuCoreFreqOperator=more_or_equal&gpuCoreFreq=&gpuShaderFreqOperator=more_or_equal&gpuShaderFreq=&gpuMemFreqOperator=more_or_equal&gpuMemFreq=&numberOfVideocards=&gpuSocketId=&gpuCoolingId=&mbModel=&mbModelId=&mbChipset=&mbChipsetId=&mbManufacturer=&mbManufacturerId=&mbChipsetBatch=&mbChipsetBatchId=&mbCoolingId=&memType=&memTypeId=&memProduct=&memProductId=&memManufacturer=&memManufacturerId=&memBatch=&memBatchId=&memFreqOperator=more_or_equal&memFreq=&memTCasOperator=more_or_equal&memTCas=&memCoolingId=&psuWattOperator=more_or_equal&psuWatt=&psuProduct=&psuProductId=&psuManufacturer=&psuManufacturerId=&diskCapacityOperator=more_or_equal&diskCapacity=&diskProduct=&diskProductId=&diskManufacturer=&diskManufacturerId=&dateAfter=&dateBefore=&_bestHardware=on&orderBy=score&limit=20&_imageAttached=on&_videoAttached=on&offset=0[/noparse]
With that query, both piledriver and bulldozer are shown:
-
Yes, it is. You need ~300-400MHz to compensate the bad performance.
Thats about right in my book.
- At 7.81GHz, Bulldozer could do 11s flat (really good efficiency). Massman is running 180MHz faster, yet scoring 281ms slower.
- I did 11.076s@7.782GHz on FX-8120 (not great efficiency). Massman is running 200MHz faster than me, but .2s slower.
- On Bulldozer, there are only 4 SP1M submissions above 7.9GHz, but all four are under 10.8s: http://url.hwbot.org/10PpQZt
So basically, at the same frequency of 7.9-8GHz, Bulldozer was about a half second faster than Piledriver in SP1M. With the right efficiency, 11.281s sp1m should be doable on bulldozer around 7.6GHz.
Really though, hardly anyone bothers running sp1m or sp32m on bulldozer or piledriver under LN2. No one has even ran sp32m on Piledriver above 7GHz.
-
Good stuff Dejo!
-
Adding fix to first post.
For a limited time, please contact me for help reproducing the fix. (I'd just like to confirm it works for anyone other than me)
So far, in my testing with the same settings and a 9800 GTX+, it took my score from 378k with wrapper to 440k with wrapper. All along, running aquamark directly would give me 440k.
-
Found a workaround for the low score with old wrapper bug. Join Overclockers benching team to get it.
I kid. Happy New Year...
When the wrapper launches the benchmark it launches a couple additional tasks you can view in task manager. Set these additional tasks to one core affinity and low priority. Score through wrapper is then equal to running aquamark directly. Alt+Tab out of aquamark to enter task manager and apply affinity/priority on other tasks.
-
Requiring recordings is a bad idea. I can't keep video records straight for every session. Let alone doing subzero, where often times results aren't repeatable - most any score I submit is the product of dozens of runs which I wasn't happy with submitting, then only sending in the result I was happy with. Often times my best score comes from somewhere in the middle - working my way to tighten down all settings to maximize the score, then spending a while pushing harder and failing to get any better. I'm not sorting through a 3-9 hour benching session to extract the 25 minutes worth of recorded results where I wasn't failing. For instance, I spent about 3 hours on LN2 getting these results which weren't that hard but just took a lot of trial and error to find what the chip's limits were:
http://hwbot.org/submission/2336586_
http://hwbot.org/submission/2336584_
HWBot isn't successful because its the perfect system. It's successful because its the best system - its convenient, focused on community driven teams, reasonably well moderated in most categories, and caters to a diverse enough audience.
Perfect is the enemy of completed. In some things, good enough is ideal, and judging from the way hwbot operates I think they often times get that balance right.
I.M.O.G. - Radeon HD 4890 @ 1140/1200MHz - 47723 marks 3DMark05
in Result Discussions
Posted · Edited by I.M.O.G.
Yes and no. I used vista, xp, and win7 to get this score - 7 was just the one that sort of worked last night. Vista32 and xp typically produce the best results with this card, but my vista64 install was bugging, I couldn't find my v32 disk, and xp was producing bad scores.
Win7 was also producing bad scores compared to my previous v32 subs on this card, but it was close enough to good that extra volts and MHz got me a decent score. This would probably be a 48k score on a proper OS.