|
@anandshimpi posted:anandshimpi
|
# ? Oct 12, 2011 18:51 |
|
|
# ? Dec 10, 2024 09:42 |
|
movax posted:This is going to be turtles all the way down isn't it.
|
# ? Oct 12, 2011 18:55 |
|
rscott posted:Jesus, this is a blunder on the level of Netburst for AMD, and I don't believe they can afford to make a mistake like that. Hopefully the graphics division can keep the company afloat long enough for AMD to either work the kinks out of the process or pull their heads out of their rear end and deliver a product that doesn't have the IPC efficiency of 8 year old parts. Given that the cpu has APU/Stream processors I wouldn't be surprised if bitcoin miners attempt to prop up sales by buying the cpus. Turns out they're waiting for the next iteration of the cpu. https://bitcointalk.org/index.php?topic=26145.0 Devian666 fucked around with this message at 19:33 on Oct 12, 2011 |
# ? Oct 12, 2011 19:30 |
|
BD isn't a APU. It has no integrated GPU so no stream processors. That isn't coming until mid next year or so. Also updated BD is supposed to be around 10% faster according to the rumor mill. If that is true then there is no point waiting around for Piledriver either.
|
# ? Oct 12, 2011 20:09 |
|
Looking at the hardwareheaven.com benches it seems they just used insane settings so that all the games were GPU-bound - like its a graphics test or something. After that its just a matter of running each test enough times and throwing out the high scores for Intel and the low scores for AMD. Had an 8150+Mobo in my cart on Newegg and plenty of time to finish the check-out but could not bring myself to do it. $100 more than the Phenom II x6 1100T AND I need a new motherboard? Hmmmm. I still want one though - will wait for FX-8170. Also I love the little A8-3850 in my HTPC - so what AMD lacks in the low-volume enthusiast gamer market perhaps can be made-up in the "I want the cheapest computer you have that plays games" segment at BestBuy.
|
# ? Oct 12, 2011 21:07 |
|
A swedish site got it to run at 4.9ghz, guess the power draw? 490. About as much as my 5970
|
# ? Oct 12, 2011 21:13 |
|
The i3 is a really good processor, though, is AMD genuinely competitive in the region where you can get a Dell with an i3 for like $450?
|
# ? Oct 12, 2011 21:14 |
|
roadhead posted:Also I love the little A8-3850 in my HTPC - so what AMD lacks in the low-volume enthusiast gamer market perhaps can be made-up in the "I want the cheapest computer you have that plays games" segment at BestBuy. A prebuilt Core i3 is that already.
|
# ? Oct 12, 2011 21:18 |
|
I have a AM2+ board and 4 gigs of DDR2...I was going to go BD with my next build, but I think the cheaper option is to just drop a Thuban in here for now... Since the only game I plan on playing takes advantage of as many cores as possible, it seems sensible (BF3)
|
# ? Oct 12, 2011 21:33 |
|
Dradien posted:I have a AM2+ board and 4 gigs of DDR2...I was going to go BD with my next build, but I think the cheaper option is to just drop a Thuban in here for now... If your mb supports it, it's not a bad way to get some more time out of an aging build.
|
# ? Oct 12, 2011 22:42 |
|
trandorian posted:A prebuilt Core i3 is that already. That A8's APU's going to blow the HD3000 out of the water. A prebuilt with a video card in it'll beat it, but I think the only place AMD's got a chance right now's if they're competing with the HD3000. Which means if the Ivy Bridge chipset's better than the 6520G, I'm out of ways to even defend AMD.
|
# ? Oct 12, 2011 23:03 |
|
Hog Butcher posted:That A8's APU's going to blow the HD3000 out of the water. A prebuilt with a video card in it'll beat it, but I think the only place AMD's got a chance right now's if they're competing with the HD3000. Prebuilts come with video cards. So yea, AMD still screwed.
|
# ? Oct 12, 2011 23:13 |
|
Hog Butcher posted:That A8's APU's going to blow the HD3000 out of the water. A prebuilt with a video card in it'll beat it, but I think the only place AMD's got a chance right now's if they're competing with the HD3000. That A8 APU is slower than a discrete 5570 which is in turn molasses slow compared to a 5770 which I consider it the minimum GPU required for current gaming at a decent resolution, and that is NOT FAST ENOUGH for many! No gamer is going to touch it just because its 2x better than HD3000. Besides the CPU portion is subpar to boot.
|
# ? Oct 12, 2011 23:28 |
|
trandorian posted:Prebuilts come with video cards. So yea, AMD still screwed.
|
# ? Oct 13, 2011 00:26 |
|
Alereon posted:Not anymore they don't, the low-end discrete videocard market basically died a few months back. Shipment volumes fell off a cliff, only high-end and mid-range cards actually find their way into systems today. It's also important to remember that the Llano platform is substantially cheaper than the i3 platform. The most expensive Llano is only $10 more than the CHEAPEST i3, and the next model under that is $10 below said i3. When you factor in the graphics performance and the fact that Intel couldn't write a working graphics driver to save their company, Llano really is the perfect platform for OEM systems. Still, the margin on these units is far lower than it is on the "enthusiast" processors. AMD can claim to have the edge on the "bread and butter" market all they like, but with Intel having a stranglehold on the price/performance market, AMD is in for a tough year to say the least.
|
# ? Oct 13, 2011 02:04 |
|
Isn't the OEM market more profitable than the enthusiast market because of volumes?
|
# ? Oct 13, 2011 02:11 |
|
The server market is where they needed to regain some ground, having declined from 25% marketshare to 5% marketshare now, and VEEEERY clearly aren't going to do that, not with power draw where it is. And I doubt many people making supercomputers are going to be sticking these things in them either, which would be pretty bad as apparently that represents a substantial portion of the current 5% they hold.
|
# ? Oct 13, 2011 02:45 |
|
Which benchmark best represents what servers need or will do with Bulldozer? Besides the atrocious power draw figures
|
# ? Oct 13, 2011 03:29 |
|
Spiderdrake posted:Which benchmark best represents what servers need or will do with Bulldozer? It honestly depends on the workload. If all of the cores are loaded well, it could perform like the best benchmarks and have reasonable performance/watt at stock clocks (albeit not as good as Intel). If anything goes wrong, it starts sucking ducks.
|
# ? Oct 13, 2011 03:39 |
|
Yeah, basically the best case scenario is power draw that's not exactly attractive and performance that doesn't offer any incentives for switching from Intel to AMD. Or it could eat poo poo and be terrible and just a really obvious wrong choice. But the big problem is just no reason to switch, every reason to stay, even if they can squeeze out that extra 6% with the kernal patch. By the time Piledriver makes what I hope to god is a better showing than this, at least, Ivy Bridge will have already happened a while ago, bringing superior thermal performance and all that along with it, and Intel will be on to the next thing. They're really well positioned at this point, they basically need to gently caress up badly for AMD to have a shot and while they aren't infallible, they already had this problem and won't be revisiting it. I haven't read about NetBurst in forever but wikipedia talking about an architecture from 2000 reads like a find and replace descriptor of what's not to like about Bulldozer. It isn't a totally analogous situation, but it's awfully similar. Remember when Intel was talking about super high clock rates? TDP >100W for a single core processor when the higher clocked models were considered? Finally just couldn't dissipate the heat at all (despite the ridiculous turbine-like shrouds they used to have for them)? Hell, didn't AMD take the world record clock rate here from a Pentium 4? Then they brought out the Pentium M, which proved that the P6 architecture's efficient, shorter pipeline and dramatically superior thermal performance was still viable with more modern fabrication processes, and life's been good since. That was back when AMD's XP processors had a 12-step pipeline compared to the 31-step pipeline on the later Pentium 4 processors. The story of AMD and Intel sucks these days. Agreed fucked around with this message at 03:59 on Oct 13, 2011 |
# ? Oct 13, 2011 03:49 |
|
Agreed posted:Yeah, basically the best case scenario is power draw that's not exactly attractive and performance that doesn't offer any incentives for switching from Intel to AMD. Or it could eat poo poo and be terrible and just a really obvious wrong choice. But the big problem is just no reason to switch, every reason to stay, even if they can squeeze out that extra 6% with the kernal patch. By the time Piledriver makes what I hope to god is a better showing than this, at least, Ivy Bridge will have already happened a while ago, bringing superior thermal performance and all that along with it, and Intel will be on to the next thing. They're really well positioned at this point, they basically need to gently caress up badly for AMD to have a shot and while they aren't infallible, they already had this problem and won't be revisiting it. Nehalem was originally supposed to be a 10+GHz Netburst based chip. Ironically, the Core-based Nehalem that actually saw the light of day ran at 1/3 of the clock speed but I'd be willing to be was much faster than the equivalent performance of something like a Prescott scaled to 10GHz. Also didn't require an FX5800-style dustbuster cooler.
|
# ? Oct 13, 2011 05:05 |
|
I bought a Bulldozer ready motherboard just over a month ago for my Phenom2, as I managed to break my AM3 board's audio port.. It's on Ebay now. Gah. Nice board too. I worked out that even selling it at a loss, the cost of a Sandy Bridge 2500K + decent Asus mobo is cheaper then buying a Bulldozer chip. The whole wait for Bulldozer has been a total PITA (socket AM3, then AM3+, AM3+ being a dead socket, then AM3+ getting Piledriver, delays etc). Have had enough now, just waiting for 2500K to arrive. I'd have been all over Bulldozer if it just matched Phenom2x6 in everything, had 8 threads and less power use. And then I'd have been more likely to upgrade to Piledriver, too - at least I'd have an AM3+ board ready for them. I wonder what kind of chip they originally intended - Bulldozer was first roadmapped for '09. No way could it have originally intended 2BN transistors @ .45NM, it would be unfeasible. This current thing is like a Prescott in so many ways. I wonder if that was always the intention or not, I guess we'll never know. EDIT - the Anand posters commenting on buying Bulldozer because AMD need the money.. Wow, thats strange logic. If they must fund AMD in some way buy the decent products it makes such as Bobcat or Llano, or better yet just go outside and don't touch a PC for a few years. GRINDCORE MEGGIDO fucked around with this message at 12:56 on Oct 13, 2011 |
# ? Oct 13, 2011 12:07 |
|
I'm hoping against hope the thread scheduling is all whack and threads that should be sharing a module (and L2) aren't so the L3 is being used for things it shouldn't be (under normal circumstances). The only other explanation I can think up is they really did alter the integer pipelines significantly (lengthened them, I fear) from Phenom II and there will be no way a simple Windows 7 patch can save it. How could they not see that it wasn't reliably beating the x4 and x6 Phenom IIs a while back, unless they had some synthetic (simulated most likely) situations where it was?
|
# ? Oct 13, 2011 13:23 |
|
roadhead posted:I'm hoping against hope the thread scheduling is all whack and threads that should be sharing a module (and L2) aren't so the L3 is being used for things it shouldn't be (under normal circumstances). IIRC they did lengthen the int pipelines because they were talking about +30% clock speed over thuban and I don't think that's possible just from a process shrink. Of course GF's 32nm node seems like poo poo (surprise) so I'm sure they lost some of the gains they had there, but that doesn't excuse them for coming out with a uArch in 2011 with 60% less IPC than SB. I don't know how the math works out exactly but I'm willing to guess that would put it roughly at the same IPC as the original K8 uArch.
|
# ? Oct 13, 2011 13:42 |
|
wipeout posted:EDIT - the Anand posters commenting on buying Bulldozer because AMD need the money.. Wow, thats strange logic. Even before BD, AMD fanboys love to say how their CPUs were "good enough" for things like "Internet surfing" when AMD was and still is whacked to hell and back by Sandy Bridge...As if that somehow excuses AMD for making subpar chips not even worth the performance/price ratio.
|
# ? Oct 13, 2011 14:48 |
|
Well, the actual idea at the root of that is legitimate. Consumers don't benefit from a monopoly, without competition there's not much to keep a totally dominant company pushing the envelope and keeping prices sane. However, acting to prop up a really, really bad product is the panicked response to that underlying sentiment, and it isn't effective. People aren't going to act as a class to boycott a far, far superior price:performance product, the industries that purchase parts at a larger scale certainly aren't going to say "oh, dear, it seems AMD has lost a number of product generations in a row, we better put their power-hungry underperforming parts in our systems so they can build better stuff!" Intel's timing has been pitch-perfect ever since returning to the P6 architecture and widening the lanes. Core onwards has been exactly the right direction to both match and shape industry demand and larger market trends. Now, their actions during the period where their processors were poo poo are pretty despicable in my opinion, they went totally knives out to keep as much of a boot as they possibly could on AMD's neck with bundling practices and sweetheart deals with major brands that could be viewed as legitimately anti-competitive. I seem to remember a decision in AMD's favor in the Euro markets to that effect but I could be off... Still, ever since then it's been a continued push towards power efficiency. That really pays off if you do the math on the useful life of a processor; the one-time expense of buying Intel is negated by the substantial power savings. And they just keep managing to shrink and improve, shrink and improve, so their power usage keeps going down, down, down and performance keeps going up. Hell, Ivy Bridge looks like its going to be able to fit an actual quad into a laptop without killing the power efficiency or putting out stupid high heat levels. They're looking at putting out a powerful, quad-core processor that has a TDP as low as the old Pentium M that started them back down the road to (legitimate) success. That's remarkable. More to the point, that's the direction things need to go in. AMD's reversal here is totally baffling. It would be one thing if it offered performance commensurate to its power consumption. It doesn't. Dramatically in the other direction. (Edit: For their desktop and server parts, anyway. I know they've got a pretty slick competitor to Atom, which is way way in the other direction, but obviously not intended for anything but super lower power applications. I do still think Llano has promise, but Ivy Bridge is going to push them up against a wall and make them fight to keep it relevant with Intel devoting a lot more silicon on their lower end processors for their GPU than they did with Sandy Bridge; hopefully AMD can at least hold on to what it's got.) Agreed fucked around with this message at 15:58 on Oct 13, 2011 |
# ? Oct 13, 2011 15:31 |
|
Ja, agree totally. I am a total AMD fanboy for the reasons stated - Intel did do all that shady stuff and more, and they don't make it easy for AMD to add new instructions for instance / optimize compilers and suchlike. In short they suck. This is the first time when I've really needed to upgrade that an AMD processor hasn't been an upgrade on what I've got (an AMD processor). If Bulldozer had been as good as a x6 with 8 threads overall, I'd buy one, even though it would be slower then Sandy but this is inexcuseable, really. Building solid products and providing long lived sockets will help them much more, then they'll get my money everytime.
|
# ? Oct 13, 2011 16:05 |
|
I agree with you 100% wipeout, but what I wonder even beyond all of the stuff said is - how badly could AMD have set things up (win7/win8, BIOS)? Did they at all or is bulldozer just disappointing, period? I note that windows 7 and 8 don't show a substantial difference, yet read all this stuff about scheduler problems, etc, which were mentioned on the last page.
notwithoutmyanus fucked around with this message at 17:52 on Oct 13, 2011 |
# ? Oct 13, 2011 17:31 |
|
All I know about it really is what THG wrote about "core parking" http://www.tomshardware.com/reviews/fx-8150-zambezi-bulldozer-990fx,3043-23.html 11.9% best case improvement in the limited amount of tests they could run in Win8; no idea how much across they board they would be. Idle power use better; he didn't run much tests on power use sadly. Real problem is, Ivy Bridge will launch before Piledriver and Windows8, and it's pimped as having +10% performance increase. And 600 million transistors less then BD. Not good (for AMD) >is bulldozer just disappointing, period?< I think it is unless you have very specific requirements, with the manufacturing process the way it is. I'll throw a Geekparty if AMD pull an A64 with this one, though. GRINDCORE MEGGIDO fucked around with this message at 19:02 on Oct 13, 2011 |
# ? Oct 13, 2011 18:42 |
|
notwithoutmyanus posted:I note that windows 7 and 8 don't show a substantial difference, yet read all this stuff about scheduler problems, etc, which were mentioned on the last page. 10% free performance in lightly-threaded applications is pretty substantial and reasonable, all things considered. It's not like the modules are as un-core-like as Hyperthreading, which duplicates only floating point execute units. The BD modules are really really close to being two cores, so scheduler inefficiency isn't gonna hit them THAT hard.
|
# ? Oct 13, 2011 18:45 |
|
It's begun, a poor goon posted in parts megathread about perhaps getting this processor, wooed by the eight-core marketing. We were able to save him, but how many more will fall! Ok, so it's not like they're picking up a Pentium 4 vs. a Athlon 64, it'll still deliver performance, just sucking down more power than a comparable Intel and suffering from unoptimized schedulers/software.
|
# ? Oct 13, 2011 21:25 |
|
Poor bastards. But, hey, you're right, this isn't like a Pentium 4 or anything ridiculous. It's more like they're picking a quad core Pentium 4 D, but with modern instruction sets
|
# ? Oct 13, 2011 21:28 |
|
I can see a lot of people buying them and getting buyers remorse. Then again a large portion probably won't know any better. I was chatting with my nephew about this release of cpus. He said that he bought a quad core and he didn't even know why. He just had a lot of money to blow on a computer at the time. He doesn't even do video editing or other stuff that would make decent use of the quad core. AMD will get revenue for having the largest 8 core penis substitute. Despite the fact that most people never get beyond using 2 cores.
|
# ? Oct 13, 2011 23:25 |
|
Don't most people who don't know anything about computers just go with Intel because the brand?
|
# ? Oct 13, 2011 23:40 |
|
Ragingsheep posted:Don't most people who don't know anything about computers just go with Intel because the brand? Think so.. some people are even weirder. Guy I know vaguely is adamant he wants to buy a CPU for £700 or so; all he does is game. I'm guessing he means a 980X. So he equates £ = value totally, and would not consider anything else because the 980X is more expensive. It was a strange conversation, the guys totally dead set on buying a £700 processor to play Oblivion and Skyrim on I could tell it was distressing him when I mentioned Sandy Bridge so I just left it; the guys on a mission, he's going to love that processor because of cost alone and who am I to take that away from him. For all I know he might want to run POV-Ray in the background while adventuring, who knows. Only reason I mention it is I can understand trusted brands to a point, but basing a purchase on cost alone is like.. its like.. I don't know what its like but its not good. GRINDCORE MEGGIDO fucked around with this message at 00:42 on Oct 14, 2011 |
# ? Oct 14, 2011 00:32 |
|
movax posted:It's begun, a poor goon posted in parts megathread about perhaps getting this processor, wooed by the eight-core marketing. We were able to save him, but how many more will fall!
|
# ? Oct 14, 2011 00:41 |
|
Nah, people don't read the OP anyway. Those of us who sorta steward that thread will have plenty of opportunities to repeat, in whatever degree of variance we feel is suitable, the regrettable turn Bulldozer has taken, I'm sure. Any submissions for "form letter to explain to non-OP reader that AMD=nothing to see here, folks" are of course welcome.
|
# ? Oct 14, 2011 00:48 |
|
Totally agree something about they would need very specific requirements to benefit from a Bulldozer - ie, run a very heavily threaded workload almost exclusively or it is unlikely to be the best choice. And even if they do run very heavily threaded apps, they should check reviews specifically containing the apps they run. Even so, theres very few cases its better then a 2600K and many many more when its worse. A bit of my hope just died.
|
# ? Oct 14, 2011 00:54 |
|
I was really hoping this would provide the basis for a cheap and efficient but beefy diy esxi/xen/whatever platform but it's looking like the extra cores probably don't make up for the extra price/heat/weak single core performance when compared to the Phenom II X6 or I haven't had an intel chip since the pentium 2 but it might be time to change. I guess it's for the best since I shouldn't be spending money right now anyway. Hey AMD you've got a couple years before I have a good amount of disposable income again. Please make good use of that time. Thanks in advance. e: didn't realize how small of an increase the 2600 is over the 2500 especially considering the $100 increase. Galler fucked around with this message at 01:57 on Oct 14, 2011 |
# ? Oct 14, 2011 01:06 |
|
|
# ? Dec 10, 2024 09:42 |
|
We should adapt the Programming Language Checklist.code:
Alereon fucked around with this message at 01:17 on Oct 14, 2011 |
# ? Oct 14, 2011 01:10 |