I.nfraR.ed

Just to be clear - I'm talking about s.A CPUs. And yes, older CPUs can be multi-unlocked (Palomino, Thunderbird, Duron). Thorton is Barton with one half of L2 disabled and on some samples it can be unlocked and operates stable, which transforms it into full Barton core. Older chips can be unlocked even with pencil, while e.g. Palomino requires e.g. silver conductive paint. Many Thoroughbred and Barton chips are also multi-unlocked, but AMD started to hard-lock them somewhere at the "famous" 39th week of 2003. For example Semprons (T'bred-B) are locked.

See you next year for results I'm doing the same with my s.A CPUs. Currently almost done with categorizing 'em in the excel sheet. Now to the point of selecting best samples per model. There are additional things like does the cpu unlocks (multiplier, cache, etc), voltage tolerance.

Ok, so something else is wrong. Maybe external cache. I don't have much experience with this platform and I only have the Gigabyte board. Will try soon with my K6-2+ 550MHz, but not going to beat my previous score on LN2, I'm sure. Strunkenbold has the best efficiency and that's the score we should aim at. I'm not even sure how good is my score, need to test at 700MHz to see where I am in terms of efficiency.

For some platforms/tasks XP is better, for some - 2000. It depends. As you can see in the pifast charts, there are better scores at lower frequencies, but using XP. So it's either that or something else like wrong registers settings/bios/board/whatever. IMO, worth trying XP. By faster, I meant the score should be much better at these frequencies (700MHz CPU, 140 FSB/DRAM). BTW, why do you think XP can't be optimized?

kotori-san, have you tried XP? The score should be better at these frequencies.

As it had been discussed before - you don't have reliable way of verifying the results. You have to rely on existing results and a common sense what is possible and what not. But that's not reliable enough. Because of this, asking for rankings is pointless, I think, although I'd love to play too. Involving points would make it appealing for cheaters. I'm speaking more generally..about all 32M low-clock challenges, not just Haswell-E. Heck, I don't even have the platform.

Yeah, that's what I sad about cpuz validation in some of the other threads. You still have some "free" space on your screenshot (according to the rules). You can move your gpu-z up and place the mem cpuz beneath, then rearrange other windows. You would even have place for another cpuz. But the best thing really is to replace that old monitor. It's not even good for your eyes. I can't stand my old 17" CRT at 100Hz refresh rate, don't know about you...

Genuine 1090T with 6th core turned off from bios. Maybe it's something about cpu core on/off feature of Crosshair V and its bios optimized for AM3+ cpus only, but as you can see older cpuz picks it right.

Another odd thing is that cpuz identifies it as X2 B60 Tell me about reliable validation.

LOL, dude. It was more like a friendly advice. You don't need to take it so serious, it's just a hobby for most of us. Memory tab isn't even that important. It's just what the rules say and if you don't want your score to be reported, then better follow the requirements. 1 year from now, someone could report your score and it can be taken down. What if you don't have the system anymore and can't re-bench? Ofcourse, moderators don't take down submissions blindly and in your case when the score is not suspicious, they make compromises. I don't say there's anything unclear in your submission, but technically speaking, you can provide completely different cpuz validation. That's why the site requires several things on the screenshot. Don't you think this is too harsh from your side? That's what I expected you to apologize for. But you know, I'm noone, I can't force anyone to do anything, it's up to you. It just doesn't seem right to me. PS: By the way, you can cover most of the benchmark screen. Noone cares which test you have selected, as long as you got the score. If you don't select the minimum tests required, then the benchmark won't show a score. It doesn't matter if you decided to run all the additional suites, they don't add up. You can see what is marked in red (as required), other things can be covered if you don't have enough space on your monitor.

Man, you should calm down and not be against everyone accusing them in cheating and so on. It is clear you're still learning and even very experienced overclockers mess up screens sometimes. Take a clear look at PCMark Vantage rules and most specifically "Verification screenshot requirements" section here: http://hwbot.org/news/888_application_16_rules/ Then tell me what is wrong with your screenshot. When you realize you're wrong, I'd guess you should apologize as well. Don't assume you're always right

Very valuable information. Thanks.

Yep. I believe this should be moved here: http://hwbot.org/hardware/processor/k6_iii_plus_450mhz/ CPUZ says K6-III+ and 256KB L2 cache

I have 2 GA-5AX and they go up to 140 only, but maybe Asus is better. Never found cheap P5A to try. As I can see, kotori did 140 on P5A, so maybe an external PLL is required to go higher.

Amazing!

Is it just me or the score is invalid? Unfortunately I see "Error in computation".

Wow, excellent scores recently and kotori-san beat me on 1M. I knew my score is not well optimized, since I lost my best wprcedit config. Need to get back to s.7 some day.

Sorry, but I think this is invalid. I had the same and did not upload, because of the red cross. CPUZ reading just jumps for a second below 50MHz, otherwise it is above that.

True. And you both have much better scores in terms of efficiency. My run is not optimized at all and what saves me is the higher FSB. I was actually expecting someone beat me with 252MHz FSB

Yeah, I saw that validation Will have to try it myself and see, but I thought minimum is x4. Never tried actually

How did you get that multiplier? FID 0 in K10Stat should be 4x, not lower. I don't have Turion, but it should be the same, FID 0 is x4. Same applies to Benchbros submission, I don't know how he got that since minimum should be x4. Did you measure by running 1M if that multi is real or you just show (again) how vulnerable is cpuz?

x0.25 multi, no way. Next step: 0 multi

Thanks guys. The technician said the said, so it seem he knows what he's doing. I still haven't tested it, because I gave it to a teammate and he handles the process.

Hi all!

That doesn't seem right. Did you change default resolution of the benchmark?