OCLab.ru Grabs Scoop on Haswell(?): First Benchmark Performance Results
It's not a full review yet, but the guys from the Russian OCLab.ru website managed to get the first testing results from the upcoming Intel Haswell architecture. They put the CPU through four popular benchmarks for overclockers: SuperPI 1M, SuperPI 32M, PiFast, Wprime 32M and Wprime 1024M. The CPU used was a Intel Haswell B0 stepping clocked at 2.8GHz and they compared it to an Ivy Bridge clocked at 2.8GHz. The OCLab.ru benchmark results:
- Super PI 32M
- Haswell: 11 minutes 27.505 seconds
- Ivy Bridge: 11 minutes 49.094 seconds
- Super PI 1M
- Haswell: 14,1 seconds
- Ivy Bridge: 13,1 seconds
- PiFast
- Haswell: 24,01 seconds
- Ivy Bridge: 25,5 seconds
- Wprime 32M
- Haswell: 12.8x seconds
- Ivy Bridge: 13.97 seconds
- Wprime 1024M
- Haswell: 7 minutes 11.181 seconds
- Ivy Bridge: 7 minutes 11.8xx seconds
Of course we were unable to verify if the presented performance results are indeed correct and to what extend the BIOS was optimised for performance. Given the odd results for the SuperPI 1M benchmark, it might be possible that the BIOS is not ready yet (or that the results were mixed up). As always, we have to be careful with leaked information you can find on the internet - pagehits and views are ruling this world - but given that the number One of the HWBOT Pro OC League definitely has a solid amount of contacts in the industry (just google a bit and you'll find out) we consider this leak to be plausible at least. In any case, if true, these results are nothing more than a first indication of the Haswell performance level - we are eagerly waiting for the first information about the overclocking capabilities to leak out.
For more information, check out their news article: link.



I guess they pissed off Francois! :p
we got ivy-e samples also here in the lab :)
I don't see the point in publishing results from such early samples because the final CPUs will behave differently. Anyone know what board they're using?
good to finally see some Haswell #'s
Interesting 1M numbers...
Ya, only way 1M numbers make sense is if the single threaded integer performance is worse, but has wells memory performance is far better, making 1M slower, but memory intensive 32M faster. That's the first explanation that comes to mind, but it is possible the 1M results were just swapped, or otherwise erroneous. @Handley: Agree its nearly pointless. But after doing a CES full of nothing for our crowd, any news about haswell is interesting to me. We need cold results, if it can't do full pot, can't compete with ivy.
Thanks for the info and the leaks :D
I want to see iGPU numbers
Pics or it didn't happen!
No idea. Assuming the leak is legit (it's still unconfirmed), my best guess would be on either an Asus or an Intel one, based on previous benchmark results/contact info, but it might be anything else. Maybe it's just a mobile mainboard from a notebook vendor :p
Probably just swapped. Anything else is just too complicated as an explanation :D
Graphs are a bit unrepresentative, making the difference look bigger than it is and going 'look, A < B!'. Put them all down to zero for scale. Also would much rather see production CPU results, even at stock, but given the release date aren't we still a month or so before anyone gets them? Pre-production samples are usually just for validation and not in any way 'optimised' for those last-minute design tweaks.
Denis, stop teasing us with IB-E :)
Text says that "for unknown reason Haswell was slower in short Pi" - means the results are not swapped, it looks like they would check such things before writing "unknown reason". IMO It's perfectly believable, given Haswells different cache and memory organization.
It seems that François doesn't care :p
[QUOTE="François"]if you see some #intel #Haswell numbers online, I doubt their veracity,since they are nowhere close to what I have in my lab, please ignore[/QUOTE]
Need something modern...How looks it in x264FHD benchmark or in Cinebench R11.5 in comparison?:)
I'm just interested in whether Haswell will have a CB or not. The maximum multi is quite promising tho.
Want to see it with frost hanging off the edges ;)
I think, no CB for Haswell :)
Full pot hooray!
Okay, got a few emails in. Apparently the Wprime would be somewhat correct, but both SuperPI's are completely off. Seems like François was right. Not sure what OCLab was upto ... just made up numbers to increase traffic? :confused:
Scores might have been legit for some wild ES/validation BIOS. We all know this is a problem with leaking performance so early. The whole article was done for traffic......
The numbers don't make sense even for ES/validation BIOS ...
says:
Many pcs in china already =_=

~~ easy test validation L1 stepping
What did your full pot hooray comment mean? Was that an affirmation, or just hopeful speculation?
Definitely one of those two.
This is a confirmation that Massman's post was a confirmation :D
Massman, Do you know that www.OCLab.ru - it is IT site with news or reviews about hardware and overclocking? If we have a hot news, which we received after our tests, why we cant share our impressions? Every news in Internet was done for people and for traffic. And it`s normal. If you not sure about valid this numbers, after NDA time Smoke will send you screens, but no need this comments, that it was just for traffic, it was done just for sharing, sharing for community. We never said in article that it was on latest ES version or it was done on retail MB, we just share, what we got, but your comments seems strange. If others received another scores, please show us. :)
Why are so many details left out? If you publish stuff lacking a lot of information, it leads to more suspicion.
Thanks for sharing, and the comments. As I said before, I appreciate ANY news.
I just think with any form of journalism, the more details the better, and for an article like that the natural questions are what board, bios, and why superpi1m is out to lunch. I think the missing detail raises more questions.
https://twitter.com/FPiednoel/status/297005032992284673
Now it's a party. Oh man.... :nana:
For those dubious about the SuperPi result consistency and comparability - please note:[QUOTE=OCLab.ru]Windows 7 x64[/QUOTE]
Thank you for these numbers guys!
Did you actually read the words I typed?
"As always, we have to be careful with leaked information you can find on the internet - pagehits and views are ruling this world - but given that the number One of the HWBOT Pro OC League definitely has a solid amount of contacts in the industry (just google a bit and you'll find out) we consider this leak to be plausible at least."
All I'm saying is that we have to be careful with what we find on the internet. Francois (and other sources) have already mentioned that your numbers are not entirely correct - possibly due to B0 stepping or OS used.
If some idiots speak that our tests wrong please make Video with same CPU, same MB, same OS, same bios options and we will see the same numbers in tests!
Guys, can you post some CPU-Z screen? With hired some information?
I don't think anyone is saying the tests were done in a "bad" way, but if there are weird performance numbers that's certainly one of the most likely reasons. Wouldn't be the first time that happened.
We need a legitimate bulletproof source to confirm this. Where is OBR?
'Fixing' the charts.
I got response only from Francois
wahts i know, results are correct for clock-to-clock comparision, BUT .... :)
LMAO ::eek:: There he is!
But... Will it blend?
:P
ty and nice, nice :)!
Please log in or register to comment.