Please can Bit-Tech upgrade the software used in the Media Benchmarks? I'm not saying this because I have a problem with my score, but because significant parts of the software have been vastly improved since the benchmarks were originally written, and so can stress multicore systems better (especially handbrake and 7-zip).
Many people use that software, it makes for good real world measurements. You could always go by SuperPi or Whetstone or whatever, but those don't mean anything in the end.
Ah yes, but they don't. What people use is newer versions of that software. For example, the latest versions of Handbrake can properly stress 4, 6, and 8 core CPUs - unlike the version used in the media benchmarks that can only seem to load my PC to 33% (roughly 66% if we take ht into account). I'm not sure what kind of speed up that would give - I don't know the settings used - but from my old computer (which is effectively the baseline 1000 pc (it gets 990) I see a 4.5x speedup for a variety of different operations - which considering that my cpu is only about 66% loaded seems about right for the extra scaling. Meanwhile, the latest versions of 7-zip support encoding up to (at least) 8 threads - making encoding massively faster than the version used in the benchmarks. This time, whilst I don't know the precise options I can get a similar result by comparing file sizes. Compressing the same files the cpc benchmark does (with AES256 compression), I get a result within 5MB of the CPC filesize within 21 seconds - this would give me a result of 8928 if I used this version of 7-zip in the CPC benchmarks, to give a comparison of how large the improvements of the last 3 years really can be. I'm not looking for an OMGEPEENLARGENUMBERS benchmark, just an update to reflect the fact that the current benchmark is not comparable to actual current uses of the same modern day applications. Many advances have been made in the last 3 years, but the current media benchmarks do not seem to reflect them so well.
totaly agree, i take it that would mean updating the baseline '1000pt' pc to something newer as-well? just out of interest what would you use for a baseline pc now if the b'marks were to be updated? something like a q9400 with 4gb ddr1066 or would keep it dual core and go with core i3/15?
Based on how it was decided last time, it would be mid-range current generation. Probably something Sandy Bridge if I were in charge. Maybe a 2500 on an H67 board.
That is kinda the point of the benchmarks though... to show improvements of tomorrow's technology over todays...
I totally agree with you. Something like; Asus P8P67 i5-2500K 4Gb 12800 Gelid Tranquilo MSi GTX 560 Ti Antec TruePower 650 SpinPoint F3 1Tb
I'd agree, the current reference platform is two generations old -- didn't the benchmarks used to have a leaderboard function as well? It'd be nice to have that back. I heard ruminations a while back a new version of the benchmarks was in the pipeline, but haven't heard anything about it for a while -- what do the team say?
We're a bit snowed under right now working on another special project (you'll know what I mean in about 7 weeks time ), but after then it is my intention to get the benchmarks updated (and yes, I know I am guilty of saying that last year too ).
ooo, you have my interest. Offsetting us by 7 weeks is a sly but brilliant tactic. I'm looking forward to this announcement.