So I've moaned about this for ages. Come on folks, we need to see some more CPU benchmarking please! Until this is all nicely resolved I wont be able to respect you Bit Tech, I feel about you the same way I feel about myself after a seedy drunken fumble behind the BMW garage. Slightly ashamed and embarassed. Here is a list of processer which need to be reviewed ASAP. Some of these were almost sort of covered in the "Best CPU Under £100" article, but that's not good enough damn it! Various socket FM1. There's only the 3850 at the moment, ZIS EES NOT ACCEPTABLE! NEIN NEIN NEIN NEIN NEIN! G620/850. Hudson/Brazos/E-350/Zacate... Whatever the hell it's called, it needs reviewing with a media/file server bias in mind. The other Bulldozer CPUs, they might be rubbish, but prices are coming down enough now so that someone somewhere might actually want to use one. Please resolve this issue ASAP so I can set my phasers to love you.
Heres a comment to rile someone: Learn how to review CPUs. BD sucks in single threaded apps? ok, true. What these days is single threaded? Don't blame AMD because Windows can't multithread its way out of a wet paper bag. Check the Phoronix reviews of Bulldozer, it shows how badly Windows struggles with any new CPU architecture!
Sadly it seems than many of AMDs A8 and A6 series CPUs have gone the way of the dodo due to yield issues. We do have another in the works though. We've already reviewed the G850 and G620 with our usual suit of benchmarks...
I agree with what you're saying, but in the meantime you always have this: http://www.anandtech.com/bench/CPU/2
Apologies, I mentioned I'd see the Pentiums in the sub £100 CPU but had not realised they were actually fully bench marked as well. I just didn't get on with that article... Oh I know. I suppose it's something that BT has to think about, it can never really compete with the detail that AAT go into and the facilities availible there but I think to actually be credible, there needs to be a clean sweep of what's availible in the CPU market currently.
I have to agree with Confusis. Reviewing a CPU needs to cover all the bases. Sure BD sux balls at single threaded but c'mon guys, how long has multi threaded and n-threaded been about, can we really blame AMD for this (i'm an Intel player to the core, ask confusis, but fair is fair). Also, whilst windows is the OS of choice for (i guess) 99% of users here, in situations like Bulldozer where the CPU scheduler patches have already been merged into the mainline linux kernel and windows 7 still has no clue about the shared front end of the "cores" (or "modules" as they seem to be known unofficially) would it not make sense to at least run a few benches on that OS as well, to show the capabilities of the CPU on an OS that actually has the needed code to handle it. Sure in some areas BD is behind intel, in others it p****s all over them under linux, its a very different game to how it is on windows and shows what the CPU is capable of under an OS that has code to support it. If your going to review / bench things, u have a responsibility to be both unbiased and comprehensive IMO. my $0.02
But how can a small team of writers test a whole field of CPU's so extensively? Testing them on multiple OS', one of which has been modified to unlock the strengths of a particular type, seems counter-intuitive to me. You're always reviewing for the masses, and even in a community like this the overwhelming majority will be using Windows 7 for their intensive tasks (i.e. gaming and CAD work). As a reviewer you have a responsibility to try and represent the facts in an unbiased fashion, but you don't have to be completely comprehensive in your testing environments - after all, it's down to the end reader to decide if the testing methods fit how they think it should be. For me the reviews are spot on - they reflect typical usage that I put my system through, I'm not going to switch to Linux for a chance on Bulldozer being fast, I'll take Sandybridge and the OS I've got going now.
I agree to a point, you guys are, ultimately, writing for your target audience, a wise move, given that its that community that, ultimately, brings value for vendors, advertisers, the rest of the community, etc. *But* i do think that at least a mention of some issues is worth while when, in the case of bulldozer especially, there are blatant software issues with the platform (OS) being used to perform the testing, reviewing, benching, etc. Sure, the majority of people reading that article may be on that particular OS but even so, there is an issue with the implementation that is not *technically* all down to the hardware at hand. If your going to look at the strengths and weaknesses of a piece of hardware then would it not make sense to at least research what those are throughout said hardware's target user base and make some sort of mention of them, perhaps with links where appropriate. I guess by "comprehensive" i don't necessarily expect a small team of writers to test for every possible corner case that might just happen to come across some review or article posted to bit tech (that would be madness) but i would expect that if there are problems on one OS and not an other for example, or other such issues with hardware (CPU's in this case) that there would be mention of these issues in an unbiased manner and perhaps one or two additional tests / benchmarks or some such to show this, i do not think allot would be needed but just a small amount of additional detail. at the risk of being unpopular, i would say "especially in a community like this" which is why i would not expect to see news on the latest developments in mesa here, there are other sites i would frequent that serve such niches. Just like i would not expect those sites to posting news about how overclockable the latest edition of <insert your favorite GPU make and model here> is. As always, my $0.02, i dont expect anyone else to share my view or change their behavior because of it but ye, there you have it.
Just worth point out, me != staff. All opinions are mine, and nothing to do with the editorial team. I agree that info, such as that Bulldozer is limited by Windows, should be included in a review if it's available at the time of writing (and isn't hidden away on some obscure message board somewhere, you can't expect the whole internet to be scoured for each review), but in the end you have to accept that most, if not all reviews are done for quite a wide audience and will reflect a broad range of people's needs.
How is it Microsoft fault that bulldozer does not work well on windows 7, windows 7 came out before bulldozer? And the supossidly "extra performance" is this http://www.bit-tech.net/news/hardware/2012/01/12/scheduler-updates-bulldozer-boost/1 isn't it? And I don't see Intel struggling to get good performance out of windows 7. All I see is bulldozer being slower than a i7 920!
Yup those SB-E chips are seriously slow at multithreading in windows Sent from my HTC Sensation XE with Beats Audio Z715e using Tapatalk
My bad, it was late when i posted and i just didn't make the distinction between forum rank and staffer, apologies on that. I guess the definition of "obscure" is in the eye of the beholder, it depends on what sites you follow, what community's your a part of and where your interests lie. But yes, i guess, in the end, your are right, content on bit tech is being written for the widest possible audience. I think there are a couple of things to consider here. Firstly, since when has hardware being newer than Windows been an excuse for shoddy performance? Im not denying BD has some big week spots and some strengths too but i am saying that the windows development ecosystem is partly to blame. I keep coming back to linux, mostly because , in this scenario, its the polar opposite of the situation in windows. For example, AMD were able to submit their patchsets into the kernel with time to spare before the BD release. After the release they were able to just as easily submit a few optimization here and there too. Hell, even Intel tend to have support ahead of time, basic support for SNB was in kernel a good year or so before release and support for IVB has been the same. My point is, that releasing new code for a kernel and then issuing an update isnt exactly some sort of deep magic that can shift the blame elsewhere. Also, if the windows dev ecosystem wasn't still 10 years behind, developing mostly single threaded apps then perhaps things would, again, look different. Alot of speculation on my part but i do feel part of the blame has to be shouldered by MS and their development policies, Windows7 really isnt a good player in the multi-threaded game.
