>>45463243 >implying AMD reverse engineers anything from intel AMD doesn't have the manpower or dollars to attempt it. Now they stick to copying obscure patents and licensing IP from strange firms no one has ever heard of before.
Not even shilling, if you want AMD to get on Intel's level then they need more money. If AMD had Intel's money they would get Intel's engineers and be making CPUs better than Intel. Theres nothing complicated here, just needs more money
>>45463531 Not yet, its unlikely that they'll release anything about it given that they haven't even really spoken about K12 yet. We still don't know the specifics of the Excavator core for that matter and those parts will start coming out early next year.
>>45463807 Their K12 and Zen lines are 14nm FinFET. The FX series has stagnated because theres no money in it for them at the moment to continue releasing marginally faster desktop chips that no one will buy. They're still selling off Vishera chips that have been sitting around in a warehouse.
>>45463840 The already released Kaveri chips use GloFo's 28nm SHP node, the Carrizo APUs will be AMD's second generation of big core 28nm chips. They've also produced Kabini/Temash and Beema/Mullins chips at 28nm. Saying that GloFo's 28nm node is far behind intel's 22nm node is just short of a bold faced lie. In terms of gate pitch they're very close, and GloFo's 28nm node produces a 25nm channel. Intel's 22nm Trigate process produces 26-27nm channels. The only key metric that GloFo's process is behind in is cache density, but this is to be expected. Intel has always been ahead of the curve here.
>>45463931 do you have high expectations for AMDs next products? or do you think they will compete closer with intel than they have this/previous generation? i know that intel is usually ahead but its a bit depressing that amds current stuff is so far behind
I wish they wouldn't focus too heavily on APUs atm but that'd what they're doing. I think they're too afraid to do anything too big as it would also come with a huge risk. Do they're playing it safe for now I suppose. Maybe next year or the year after we'll see some new chip that is actually worthwhile. Maybe they already have one in the works and they're keeping it under wraps.
>>45464085 >do you have high expectations for AMDs next products? I'd really like to believe that they'll deliver a chip thats competitive across the board, but I seriously doubt it. Best case scenario is that they become somewhat competitive in serial performance while still offering top tier IGP performance. That at least will let them compete with intel's low TDP parts, Celerons, Pentiums, i3s, and their locked i5 SKUs. They could still regain some server market share by pricing their chips well below a Xeon. Provided that power consumption is roughly equal the price/performance ratio would be in their favor.
> do you think they will compete closer with intel than they have this/previous generation? When the Zen chips are released they'll be competing against intel's Skylake architecture. If AMD's new arch was anywhere near it in serial performance and power consumption then it would be a win, even if they were 10% behind. The last of AMD's 45nm chips weren't too far behind intel's 45nm chips. They've been close before, and with dropping the whole Bulldozer derivative architecture theres no reason they can't be close again.
>>45464218 Even if it was a minimal performance boost, AMD could just do what AMD does and bolt on more cores. That would be the real breakthrough. Wait for Intel to announce a tick or tock, and the day after announce and release a 64-core desktop processor which can run multiple cores per thread.
There's no "problem" with AMD to begin with. Since the C2D was introduced Intel has been ahead at single threaded performance and if you look at benchmarks even the Bulldozer chips were realistically equal to their Intel counterparts regardless of Intel cheating and buying the press focus on single threaded ICC benchmarks.
The only reason to think AMD has an issue is if you think a 32nm chip won't be naturally "slower" than 22nm chip.
>>45464218 I have a very high suspicion that you could do that for the x86 ISA, if you could I'm pretty sure we'd be seeing at least some kind of experimental prototype from either AMD or Intel in that direction. It's probably only possible for less complicated shit like RISC
The VISC arch from this company describes a kind of clustered simultaneous multithreading. Its like hyperthreading on steroids where every core is virtual, and can work on the same threads. Though again, I don't know if this has ever seen the light of day so their claims could be greatly exaggerated. They're not the only company to develop an arch that does this either.
>>45464347 Honestly it seems logical for intel to attempt something like it given how long they've had SMT implemented. If AMD doesn't do something like it intel eventually will.
>>45464363 AMD would have to figure out how to get it to work with x86-64 though. Otherwise, we''d just end up with Itanium again. They have a whole pile of chip wizards and a femme MIT grad CEO. I believe in the new AMD.
AMD GPUs can be competitive in the future IF they manage to make proper drivers. If they don't, the GPU division will be fucked, and I bet my ass that that's what's gonna happen. Their CPU division is pretty much fucked already, so here's what I predict: When the current generation of consoles will be over, they'll stop producing x86 chips, because they'll be so much ahead of Intel that even with uber-low prices, nobody will buy their shit anymore. Then, they'll try to survive making cheap ARM processos, but it will be too late for that, and therefore AMD will go bankrupt by... 2022 or something.
fuck anything x86. Of course they would make major losses in the beginning, but maybe it could work out at some point, especially since a complete overhaul of the architecture becomes more urgent every day.
He's a COINTEL shill, the 280x I bought in August has NEVER had an issue or driver hiccup. The only issues he's referring to are the natural immature driver problems that plague AMD and nVidia equally... hence throttled $1000 Titans that perform worse than a $300 AMD.
>>45464564 you mean: what doesn't suck? I've had performance and stuttering issues since day1; issues with many games older than 2-3 years; OpenGL support is basically non existant; the first releases of that god awful driver even had problems with the fan and killed several GPUs (including mine). I've had so many fucking issues I can't even remember them all. The sad thing is, this piece of shit of a GPU is as fast as a 970, BUT a 970 runs most games 30% faster, without stuttering, and without being a fucking housefire. THIS is the worst video card I've ever had, seriously
>>45464673 Let's examine one of the issues I've encountered: OpenGL support. Basically, the only OpenGL game I can play without issues is Minecrap. Baldur's fucking gate enhanced edition showed a black screen (recently fixed); On the left, you can see Star Wars Knights of the Old Republic; and if you google a bit, you'll find out about issues with id3 and id4 based games like Quake 4 or Star Trek Elite Force. Basically 90% of OpenGL games either glitch or don't work at all. This is shit.
>>45463054 Intel is is a shameless company with no ethics. Looks up the Intel fortran compiler purposefully crippling performance on anything nonintel. They got sued by the FTC. Also they give away free processors to make their image stronger because they're rich jews and they know amd won't get used by manufacture when atoms are being given to them for free. Apple uses slave labor and shit proprietary sources and hardware.
Amd wants open standards for technology so we can all move forward . And AMD is the bad guy here?
This thread is why tech has gone no where in in the past 5 years. Fucking idiots.
>>45463067 You're joking right?Intel massively overpriced their CPUS until amd could compete forcing them to lower prices. You want an evil monopoly? You're breed is is what allows the rich and evil to maintain power . Blind sheep
>>45465245 This entire image is fucking painful to read.
>>45465292 >>45465311 Despite how retarded that picture is, what the one moron is trying to describe is actually a real thing. Its cluster simultaneous multithreading. There are a lot of white papers out there for it, for a number of different applications as a matter of fact. http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=4601890
>>45465594 Its real real. Though like SMT the performance benefit would depend on the type of workload and how predictable it was. For something like a synthetic benchmark that consistently loads a core with a heavy integer instructions then the potential performance gains are quite large. In a real world client workload however its far less certain.
I'm not into /pol/ mentality but I do think Intel is "jewing" really hard since forever. AMD got massively fucked, they can't recover and what Intel has been made to pay isn't enough imo. Plus we know Intel is just keeping AMD close enough to not be sued for monopoly, someone who's not an Intel shill should step up and fucking put some real control on them, it's getting out of hand.
We're gonna end with Intel being so rich they will be able to afford the fines from monopolists practices kinda like Microsoft did back in the day in europe.
All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.
This is a 4chan archive - all of the content originated from them. If you need IP information for a Poster - you need to contact them. This website shows only archived content.
If a post contains personal/copyrighted/illegal content you can contact me at email@example.com with that post and thread number and it will be removed as soon as possible.