the lack of PWM and Chipset cooling is worryingQuote:
Originally Posted by Fatfool
these things are designed for cases with forced airflow not a huge enthusiast case without any directed ariflow :shakes:
Printable View
the lack of PWM and Chipset cooling is worryingQuote:
Originally Posted by Fatfool
these things are designed for cases with forced airflow not a huge enthusiast case without any directed ariflow :shakes:
Show me da dozer and I shall show you both processors in their sockets.Quote:
Originally Posted by Shadov
Bulldozer is just looking too awesome (in any case, even clock for clock) to waste cash on a second processor which will be vastly outclassed
Could have used one of those enthusiast heatsinks which blow downwards like the scythe shuriken.Quote:
Originally Posted by generics_user
Good thing I swapped the stock chipset heatsink for a full copper enzotech one. Couldn't find one for the lower chipset (southbridge?) cooler though. :shakes:
I guess the circular one would work:
http://www.enzotechnology.com/images/cnb_r1_photo4.jpg
edit: nope, this one looks custom sized
No biggie, the forumers will find a way from active chipset cooling to a liquid cooled monster :ROTF:
I prefer mine silent.... hope it doesn't come back to haunt my ass in the future. :rofl:
Btw, notice the odd copper row at the top left corner of the board?
Some genius decided to forget to place a heatsink on. I'm improvising with some copper BGA heatsinks. They seem to be holding up for now but probably can't do so if I had another processor. I can imagine the look on a server admin's face when he drops in both processors without a care and it toasts itself :ROTF:
By the way, I can't get the temps on this board. There aren't any sensors whatsoever?! :shocked: Can't find it in the bios either.
Or perhaps the chipset is so new, I can't get readings on it. Anyone know of proggies that can monitor it.
Also, memtest86 throws up a ton of errors, too absurd to be true (I let it run and it kept of showing errors till it was like 2+gb and I shut it off)
Previous hot chips presentation video is now available online(HC 22,Aug 2010):
http://www.hotchips.org/archives/hc2.../session7.html
On the AMD side Mike Butler ,AMD Fellow,presented Bulldozer module and techniques AMD used to increase performance and perf./watt and decrease power and die area investment at the same time.
Interesting tidbit from the Q&A is the question about opportunistic prefetch of the cores and whether 2nd core's prefetched data in the shared L2 can be used by 1st core(answer was yes,constructive prefetch in L2 is possible). Also Mr Butler confirms 3 memory ops capability per core while there are only 2 visible AGLU units in the diagram(2 128bit loads and 1 128bit store per cycle per core). AMD was emphasizing their instruction and data prefetch and the aggressiveness of those. These features seem to be one of the major performance increasing contributors in the bulldozer module.
Lets bury for real that stupid war between overclockers and JF-AMD about 2P overclockable systems. Lets check the problem by type of potential buyers
First, you have the enthusiast gamers who build his own computer to be able to play crysis at high details. These buyers dont need 2P systems cause most of the game dont even need more than 4 core for decent gameplay and also game developpers don't want to make their game too much threaded if the majority of their potential buyer only have a 2-4 core cpu and a mid-range gpu cause that would cut their profits. So gamers dont need 2p overclockable systems
Next up, we have the extreme overclocker who can buy the best of the best and needs the best of best to stay competitive. Happily for the overclocking sport, 2p systems are bad in the benchmarks that is used in overclocking, cause if 2p systems would be doing good, it would limit again the number of people that can be competitive in this sports because the cost of the systems would rise alot. And less people who can be competitive means less fun for the top guys who when they get at the top, can never be took out from the top by a newcomer because its a very small amount of people that can buy the systems needed to get to the top and it makes it also not fun for new people to try to be competitive because the cost is way too high and they cant afford it. So the overclocking scene dont want 2p overclockable systems that can destroy overclocking world record easily.
Then we have the usual 2p systems users ( scientific, video editor, content creator and cad), happily for them, most of their normal software now use both the cpu and the gpu (cuda, mercury engine, opengl) so they don't need anymore the biggest amount of cpu power for a reasonnable cost ( a normal 2p systems cost somewhere between 5 to 10 grand while a 4p systems normally start near 15 grand) cause now they can also add some gpu power to cut down the waiting times for the work to be done( and for most of them doing more work in the same amount of time for more profits or faster research) and also mostly because they need 99% stability while gamers need about 75% stability and overclockers need 20-50% stability because they dont want to redo their work because of a unstable overclocked cpu even thought they dont need 99.999999999% stability of mission-critical 4p servers. Most of the potential buyers use their computer for their job, if the computer is not stable, they are just simply losing money. So usual 2p systems buyers dont need 2p overclockable systems.
After we have servers and as i said they need 99.99999999% stability something that a overclocked systems cant do and most of them need the most memory and the most cpu power they can get so they are mostly 4p systems potential buyers.
Finally, we have a very small group called The Crunchers, which wants the most cpu power, need about 80-85% stability, want low electricity usage, overclocking options because they dont need as much stability as the normal workstation buyer and some of them also need the most pci-e lane they can get for distributed computing projects that can use gpu powers. They are a though crowd to please and they are a small group of potential buyers, sure we would be happy to have 4p or 2p overclockable systems but we can also be very happy with 1p systems or just have the right connection for some unlocked es cpu.
So the only person that could ask for some 2p overclockable systems are crunchers and as JF-AMD said we are a too small amount of potential buyers for them to do some R&D and testing for a couple thousand of cpu sold to crunchers.
While i dont agree with some of your points (cuda isnt answer to everything).
Affordable 2P system for enthusiasts wouldnt cost that much.
That isnt changing the fact that on the Intel side, there is one ;-) .So again discussion is moot.And JF tells us also that high end single socket are useless, which is, well, only thing he can say considering AMD doesnt plan on having one.
While i appreciate JF`s many contributions to the forum and general knowledge he can bring to us.Fact is, he works for the company, so he MUST say what they expect of him and FULLY support current AMD strategy. Besides, affordable desktop 2P system is not what opteron salesman wants.
i'd have thought it is in AMD's interest to bring back the 4X4 platform , 2P desktops. even if 5% of users go for it it'll sell a lot of cpu's and make the performance difference which is massive in favor of Intel moot due to price of AMD CPU's,Quote:
Originally Posted by RaV666
Then there is always the Marketing opportunity to the "future proof" crowd.
You still dont get it ???Quote:
Originally Posted by RaV666
Desktop 2p systems with overclocking possibility with the unlocked multi that you guys ask means alot of money for testing,validation and R&D, sure its not as much as much as making a complete new cpu but its still a couple of millions dollars that amd must get back in profits. Also for the there is one on the intel side, you know that its a special one-off project made by EVGA and not Intel, its EVGA alone that decided to make a overclockable 2p xeon motherboard, i am pretty sure it has cost them a load of money in R&D and that they might still not be making any money from it but getting the R&D cost back from others revenues. Intel did nothing in this case, we were just lucky to have the possibility to overclock the bclk on those xeon and still you have to shell out 2 grand for 2 decent cpu with decent locked multi if you want to have any decent overclock. Also, yes i know cuda isnt the awnser to everything but both cuda and opengl will help alot in the future for typical workstation user as the industry standards software are starting to use them to get faster results.
And for high end single socket didnt i just proved in my last post the usefulness of them ? Overclockers, high-end gamers and crunchers all wants high end single socket, they might not be a big group alone but together they are enough to be worth the R&D.
As for affordable desktop 2p systems from opetrons, i guess you never looked at some amd 2p systems ??? a single socket lga 1366 will set you about 700$ here in canada for the cpu,mobo and memory while a 2p amd socket c32 rig could cost me about 900$, i find it quite affordable.
I believe he is saying that John is defending AMD NOT having a high end single socket system, which in many ways he is. You believe its worth having that type of system and I think everyone here in XS wants high end single socket AMD desktops. But John keeps emphasizing that will not happen any time soon. 8 core socket AM3+ is the best we will get period, and there wont be a product out to counter intel's socket 2011.Quote:
Originally Posted by snoro
I believe that is his frustration and of all of us really...
I fall into this category. Can't wait to get back to editing and encoding my videos. :DQuote:
Originally Posted by snoro
Still, At the price and performance of these C32 rigs, they could be marketed, without any substantial changes to the enthusiast (not overclocking, I mean more of DIY) crowd. (as in 'look at how this beats the 980x on price and perfromance!' and not 'zomg! look at how it can overclock and take all the benchmarks'.) after all, the boards have already been designed and produced.
but as JF has said, not profitable.
But Interlagos beating LGA 2011 :D... I am still not sure about LGA2011 and SB desktop...(Think, it will be later - Q4 2011/Q1 2012 as IB.
Actually they don't. They just need to flog them the same processors used for servers.Quote:
Originally Posted by snoro
I was talking about both the 2P systems and high performance 1P systems.In both cases i dont get it ;-).Quote:
Originally Posted by snoro
Because i just plainly dont believe that unlocking multiplier and modifying existing infrastructure (only the bios really, which im pretty sure they have internally for testing purposes anyway) would cost 5mil.
I DO get it why JF is defending his company tho.
AND 1P high performance systems (1366/2011 type of thing) which AMD isnt going for also.(Think G34 1P system for example to REALLY lower costs).
And as for the affordability, good point you can get 2P C32 system for 900$.
Which will be slower than 6Core @4.5ghz cheaper intel platform.On the other hand, if you could get these optys up to speed...
And as to the EVGA debacle, i really dont believe either intel or evga is losing money on these systems, and im sure intel had to give green light to evga, they wouldnt sell them chipsets if they didnt have the blessing.
And from what i understood from various JF comments, AMD is NOT going to give green light for such a project.
Anyhow, i just think i can summarize whole discussion.
AMD defenders say:
AMD is goin to lose money on it!
AMD attackers say:
Theres no logical reason they would!Intel ISNT!.
There is slim chance that bulldozer will be competetive even in its 8core form with 8core sandy/ivy bridge.Lets hope :up:
rav666, check how big intel is with its employers, R&D, revenues compared to amd.
come back to reality thanks
Reality check done.Quote:
Originally Posted by Tomasis
AMD did have Athlon mp boards(had one), AMD did have dual FX boards.
AMD doesnt need to do anything pretty much besides changing its approach to enthusiast market.
Check. :up:
Its companys policy and current market strategy, not inability to do so.
There won't be client 2P boards. Check.
If you want 2P ,buy 2P server class board with 2 Interlagos chips in Q3.They will automatically "overclock" with the new Turbo,so you don't need to OC it manually. You'll need a couple of grands though.
Don't get interlagos darn it.Quote:
Originally Posted by informal
Just get Valencia. Should be much cheaper.
Well he seems like uber enthusiast. Why stop at 2x8 cores,he should go all the way to 32cores(if he is that extreme) :p:
Actually what i would like the best is single processor interlagos platform that i can overclock :up:
Should serve me for 3+ years just like my AM2+ platform is :p:
Quote:
Originally Posted by RaV666
yes and it didnt turn out nice right .. and since then the recession and all ... how's that for a reality check ....
and amd doesnt have the same R&D budget that intel has ... want more ideas on why its pointless to argue about what jf-amd tells us ...
every project needs to be looked at from a financial standpoint .. R&D and ROI ... and everything ..
Quote:
Originally Posted by RaV666
cant wait till amd releases a platform with 12 cores in a couple of years ???
what are the programs you are running that really need that much cores ...
maybe example superpi :D
And now they have quad and up chips and the market for those dual systems went away. You will not get what you are wanting from AMD, it makes no sense at this point.Quote:
Originally Posted by RaV666
Main problem with using available systems is low clocks.And high cost of a LITTLE faster chips.Quote:
Originally Posted by MrMojoZ
AMD offerings for some time now are non overclockable and have low powerformance (because of low stock clocks) in any but server applications, which limits its use a lot.
Who would buy such systems ?
Well i would buy high performance system from AMD, i would build a high performance workstation also for a graphics designer i work with.
There is market ,albeit smaller than it was.
Intel and apple proves that there is.
I pretty much said all i could, and put all my points some time ago in this and other threads.
On a side note, what i find worrying/weird is the fact that people seem to "defend" corporations when its actions are pretty much set against our (the enthusiast community) needs.
Once we the consumer demanded, and we often have gotten it (xtreme editions, FX series, BE editions and now K editions) are all a product of "we want more" mentality.
Yet now i see people defending the right of intel to screw us with sandy bridge anti overclocking design ,with disabling functionality (HT VT and so on) .Yea, they can, its their product.But to actually defend that aproach when youre part of a community that needs the opposite is absurd (for me at least).
Same goes of course for AMD "fans" , i say "fans" because i think that a true fan wants his team to excel and do extraordinary things ,not be mediocre.
We the consumers, and enthusiasts, should demand more, not less.We may be relatively small .but i really dont think our impact is small, i alone am responsible for hardware buys and configuration of dozens of people, and everyone of us is.And more, people here are more often than not people who make decisions in larger corporations.
You need one board, with one unlocked bios, with a one unlocked SKU to make many DIYers happy.
Thats politics, not engineering or money issue.
first things first, AMD is smaller than Intel, why this is important:
lets say that 2P/MCM overclockable workstations are 1% of the market, and an analyst did the math and found out making a system to fit that will only be profitable if the own more than 40% of that market. (bulls**t numbers for the sake of an example). right now the market share for AMD is much smaller, and its a very high risk to introduce something that is not going to profit unless there is a MAJOR jump in brands.
secondly there is still evolution in how servers are being used.
the main way to break down needs is a PC will fall into one of 2 groups. the most powerful machine you can afford that will not be used all the time, or the most power effecient machine that will cost the least over a long period of time. gamers clearly want high power because their electric bill will be 5% of their total cost, and a cruncher might be ok with a 2000$ 2P that saves them 300$ a month compared to client systems.
But what about the people who need more power than a desktop can offer, but dont need to use it every second of the day? those needs are starting to be filled by cloud computing and on-demand systems rather than forking out the cash for an excessive machine.
5 years ago, things were 5x different. what worked then may not work today, and todays solutions will be laughable in a few more years. were seeing gpu computing coming close to replacing multicore cpus, and i think were going to care ALOT more about single threaded performance with cpus and be capped around 4 cores for client desktop as the most optimal answer in cost/benefit for a very long time, with opencl offsetting anything parallel to the gpus.
If these WILL be FX branded, don't count it out. I wouldn't mind seeing a Classified SR-2 type board out for AMD.Quote:
Originally Posted by informal
Im curios, if we will see BE and FX series together, or "only" FX serie. Hope, some guys here will have Interlagos as homestation :-)
IMO now we will see FX as Intel "X" equivilents, and BE will stick around as "K" equals.Quote:
Originally Posted by FlanK3r
Hopefully they dont overclock as bad as K CPU's, if I'm right.
...talk about beating a dead horse.....wow.
AMD released both Athlon MP and Athlon FX with dual sockets. Both failed, and that was in a time when these things were much more needed than today. Back then you had single cores, a dual socket solution made a huge difference, still it failed. How many will bother when you can get 8 cores in one single socket?
AMD made the math.
Come on dude ! Give a try at K series from Intel ... You'll see if they overclock bad ... :ROTF:Quote:
Originally Posted by BeepBeep2
Ive already responded to this, but what the hell, last one time ;-).Quote:
Originally Posted by -Boris-
Athlon MP was AMDs first attempt at multiprocessing and workstation/server market.And it did OK, boards were relatively cheap, you could mod mobile bartons to work on these.And platform was available for some time.
It did OK all things considered.
Dual FX failed, simply because AMD was selling ultra expensive platform with two power hungry cpus ,that still was slower than single intel desktop quadcore cpu, and more over, AMD and its partners pretty quickly abandoned the platform altogether.
Core2quad`s bulldozed it.
So if bulldozer IS NOT a failure ,than this analogy doesnt fit.
bulldozer has turbo.. so it doesnt need OC even in server environment
good luck buying 4 socket mobo, rav666 for your "graphics design" needs.
i work with it too and I think single bulldozer covers the needs with turbo. there arent many apps optimized for more cores.
better i add more ram, dual SSD, more hdd, triple 24"/30" IPS monitors etc.
to have server for graphics needs, it is like mount a 10 liter v10 motor on old Toyota Scarlet.
I tend to believe this is a 16 core server bulldozer LOL.
http://scarletwhore.com/?p=3277
Looks like they're against the law right?:rofl:Quote:
Well after months of begging, threats and attempted bribes we finally persuaded our AMD affiliated source to provide us with some baseline comparisons for the soon to be released AMD Bulldozer CPU.
http://scarletwhore.com/wp-content/u...er-CPUmark.jpg
First all those compared are desktop chips so Bulldozer in that chart should be a Zambezi 8 core part. Second of all ,don't believe in fairy tales from dubious sources... That "website" has already claimed that 8C Bulldozer is almost 2x faster than 980x in "3D rendering" (whatever they mean by that). So until we see some real data all leaked benchmarks from dubious sources are not to be trusted.
Highly dubious indeed. Considering that a module is slightly larger than a SB core, that would mean AMD is able to extract twice the performance from the same die area (roughly speaking). :/ I think not.
AMD actually claims they manage to get almost 2 cores worth of performance from a monolithic dual thread capable module(as they call it;another term is optimized dual core). This is achieved via clever sharing of resources. The thing is that with shared front end,all the benefits of SMT (filling in the pipeline bubbles) are still present with AMD's approach.Only this time,you don't share the same execution unit -you get a whole execution core ready for the thread. This is described by AMD as "smoothing of inefficient/bursty usage".AMD invested heavily in both instruction and data prefetch which are now order of magnitude better than in 10h family. So you are left with fully featured cores that have all the advantages of SMT and behave/perform nearly at the level of independent cores in hypothetical non-shared module(as if they were made a la Athlon/C2D with non shared parts except maybe the L2 cache).Quote:
Originally Posted by ohnoitseddy
What I'm trying to say is that with Bulldozer,we can't use the same comparison methods we used in the past in which we compare die area of single core ,say, Athlon and Nehalem,and derive the die area investments both firms made in terms of logic and cache. Since Bulldozer is organized the way it is,we have no clue how much die area would ,now shared parts ,occupy in a hypothetical non-shared design. There are some numbers being thrown around (from 15 all the way up to 30% bigger module),but the thing is that only AMD knows exact figures. Now ,since sharing has some benefits as previously mentioned and as well some possible performance downsides,AMD invested in areas which will maximize benefits and minimize the bad effects of sharing. How well they did this will make or brake the module approach and whole Bulldozer idea.
Whether Zambezi can be 2x faster than 980x in some cases I don't doubt.In optimized FMA4/AVX applications difference can be even higher than that. But in legacy code I'm skeptical. The fact is they don't need to be 2x faster,they just need to be faster ;). If they are,this means Zambezi's cores are much closer to the level of Nehalem ,no matter how they achieved it(whether it's just IPC or just clock or combination of the two). Next year will bring improved Bulldozer in the form of 10 core Komodo part for desktop. That's 25% more cores,with possible minor IPC tweaks and 25% more TDP headroom for cores to boost clocks in poorly threaded workloads ;).
Wait, what ? dubious sources, say it ain't so! :rofl:Quote:
Originally Posted by informal
Everyone knows the bulldozer will do this:
There you have it, exclusive ASCII benchmarks!Quote:
. : |
Do you have a source for that? Not that I find it impossible, but I like to work with verified info. :)Quote:
Originally Posted by informal
On SMT, I suspect Bulldozer doesn't really get all the benefits from it. SMT is used to justify creating a very broad execution engine, which would be wanted to maximise single threaded performance. It might be so (and of course we can't be sure) that AMD had to do concessions in this regard (e.g. for efficiency reasons - execution units do take a lot of energy if kept running despite their small size, right? It's a bit doubtful that AMD chose to implement gating parts of a module.). If that's the case it may be 'smoother', and will rock for heavily threaded apps, but a full-on SMT approach might still have been better for single threaded apps.
I do believe AMD made the right choice, by the way - the module is probably better suited for server workloads. And regarding the client space, I suspect that especially in the future most times the CPU is a bottleneck will be in multithreaded situations.
And of course they needn't be twice as fast. :) Since it's probably about 40 % larger than a SB quad though, it would be nice if they could sell it for at least some 50% more (given lower yields et al), to have similar margins. The "more cores == higher single threaded frequency" thought is a nice one though. ;)
Anyway, I was just commenting on the graph. I'd find it fairly absurd that in billions upon billions of dollars worth of research, Intel missed just that one thing that could double their performance per mm^2. :P Just to express my skepticism about the graph - not so much the Bulldozer architecture.
Everything can be found here:Quote:
Originally Posted by ohnoitseddy
http://www.hotchips.org/conference-a.../hot-chips-22/
It's a video from HC22.Mike Butler presented Bulldozer design and there was Q&A after the presentation.You can find all about aggressiveness(he couldn't emphasize it more) in the presentation and some of it in the Q&A.
SMT does zero for single threaded workloads so I don't know what you mean by this. In Bulldozer the shared front end is doing all the work for instruction dispatching and the integer schedulers do the actual work according to the dispatchers orders. FP unit in Bulldozer module is full on SMT approach due to latency tolerant nature of instructions it deals with.L2 is shared dynamically by 2 running threads(so SMT in essence as AMD describes it in one of the slides). Everything else is in the module is of vertical MT organization(switching back and forth between the threads).Quote:
If that's the case it may be 'smoother', and will rock for heavily threaded apps, but a full-on SMT approach might still have been better for single threaded apps.
So while integer cores do have common L1 instruction cache and front end ,they are fully independent.They do can do opportunistic prefecth in the L2 and the data can be then used by either of the cores. FP unit is dedicated or shared. Dedicated it can assign a whole 256bit FMA to one core,or it is SMT organized and shared by 2 cores as 2 x 128bit FMA(being able to even do 2x ADD or 2x MUL,a feat not possible by any of today's x86 cores).
Intel is suffering from "not invented here" syndrome. They don't like using ideas that are developed by their competitors unless they really have to(think AMD64 ISA).They had some of the people behind "CMT" approach working in intel back in the day,ie Andy Glew,but they never really backed that idea up. Since Glew moved to AMD he presented them with the same concept.Coincidentally some of AMD folk were looking into the similar direction and they decided to pursue CMT approach ,roughly in 2005. 1 year before that Glew left AMD .Chuck Moore held a presentation back in 2005 describing CMT as the best choice for next gen. of multithreaded MPUs from perf./watt/mm^2 perspective.Also Fred Webber hinted back in 2005 where AMD was heading with their next gen of multithreaded CPUs.Quote:
I do believe AMD made the right choice, by the way - the module is probably better suited for server workloads. And regarding the client space, I suspect that especially in the future most times the CPU is a bottleneck will be in multithreaded situations.
And of course they needn't be twice as fast. :) Since it's probably about 40 % larger than a SB quad though, it would be nice if they could sell it for at least some 50% more (given lower yields et al), to have similar margins. The "more cores == higher single threaded frequency" thought is a nice one though. ;)
Anyway, I was just commenting on the graph. I'd find it fairly absurd that in billions upon billions of dollars worth of research, Intel missed just that one thing that could double their performance per mm^2. :P Just to express my skepticism about the graph - not so much the Bulldozer architecture.
That's very interresting. I've had a feeling that AMDs prefetchers had a large part in their lower IPC.Quote:
Originally Posted by informal
Anyone know how large performance impact greatly enhanced prefetchers might have?
Totally legit review is totally legit :p:Quote:
Originally Posted by undone
hehe, nice big fake :)
Great, thanks!Quote:
Originally Posted by informal
Regarding SMT and single threaded improvements... SMT in itself indeed does nothing to improve single threaded performance. You must look at the whole design process and the decisions made. If you look at the Nehalem case, some strategy planning occurred first: if I recall correctly, the question was whether to do relatively narrow cores, which would be very suitable for server loads, focus all of their attention on making a broad, single thread oriented core or do a broad one that utilises SMT to make the efficiency acceptable for servers (whether or not to use SMT was indeed a question; the chip designer admitted it's quite hard to pull off SMT, so it would require a lot of resources (finances and time)). They chose the latter, so that their architecture would be suitable for both server and client. Servers have high margins - if it weren't for SMT, they couldn't have justified building a very broad core, capable of handling single threaded so well.Quote:
SMT does zero for single threaded workloads so I don't know what you mean by this. In Bulldozer the shared front end is doing all the work for instruction dispatching and the integer schedulers do the actual work according to the dispatchers orders. FP unit in Bulldozer module is full on SMT approach due to latency tolerant nature of instructions it deals with.L2 is shared dynamically by 2 running threads(so SMT in essence as AMD describes it in one of the slides). Everything else is in the module is of vertical MT organization(switching back and forth between the threads).
So while integer cores do have common L1 instruction cache and front end ,they are fully independent.They do can do opportunistic prefecth in the L2 and the data can be then used by either of the cores. FP unit is dedicated or shared. Dedicated it can assign a whole 256bit FMA to one core,or it is SMT organized and shared by 2 cores as 2 x 128bit FMA(being able to even do 2x ADD or 2x MUL,a feat not possible by any of today's x86 cores).
In the case of BD, AMD could have invested the transistor budget of the second INT cluster in making the design broader, and implementing SMT to make it more efficient in multithreaded environments. You could end up with a core / module of the same size, but better for singlethreaded workloads, though with the downsides you already mentioned.
I'm not sure whether Intel suffers that. The integrated memory controller, HyperTransport-like bus, even the three-layer cache structure all were implemented by AMD first, and Intel later adopted them. One could say the Core 2 architecture was inspired by thinking that lead to Athlon too - a relatively short pipeline and a focus on IPC rather than frequency. In fact, all this lead Tom's Hardware to title their Nehalem review "Architecture by AMD?" ( http://www.tomshardware.co.uk/Intel-...iew-31375.html ).Quote:
Intel is suffering from "not invented here" syndrome. They don't like using ideas that are developed by their competitors unless they really have to(think AMD64 ISA).They had some of the people behind "CMT" approach working in intel back in the day,ie Andy Glew,but they never really backed that idea up. Since Glew moved to AMD he presented them with the same concept.Coincidentally some of AMD folk were looking into the similar direction and they decided to pursue CMT approach ,roughly in 2005. 1 year before that Glew left AMD .Chuck Moore held a presentation back in 2005 describing CMT as the best choice for next gen. of multithreaded MPUs from perf./watt/mm^2 perspective.Also Fred Webber hinted back in 2005 where AMD was heading with their next gen of multithreaded CPUs.
If there's a reason why Intel chose not to implement CMT, I'd guess it's because Intel never separated INT and FP the way AMD did and still does, so they can't do the FlexFP trick as easily. This would reduce the gains in area efficiency, I suppose.
I do dare to guess that we'll see some form of CMT going on with Haswell... ;)
They assure this *fake* is true? WTF?:shrug::confused:
http://scarletwhore.com/?p=3277
ATI-Forum quotesQuote:
This benchmark has produced a lot of interest and controversy. I can assure it is correct. Back in November I predicted that the soon to be released Panasonic GH2 camera was going to have image sensor problems and this would cause a delay in the shipping and availability of the camera. I was correct. Beyond the benchmarks I have made 2 other predictions: Apple is going to use the AMD Fusion platform in a big way and AMD most certainly is going to release a Dual-socket Bulldozer configuration.
http://news.ati-forum.de/index.php/n...ks-aufgetaucht
If he's right, then it must be the 16 core server part. That'll compare to Gulftown & SB more or less in the way the graph describes, I suppose, since the frequency of those chips will be a bit lower than the 8 core ones.Quote:
Originally Posted by undone
Remains to be seen.Quote:
This benchmark has produced a lot of interest and controversy. I can assure it is correct.
Source?Quote:
Back in November I predicted that the soon to be released Panasonic GH2 camera was going to have image sensor problems and this would cause a delay in the shipping and availability of the camera. I was correct.
Wouldn't be surprised.Quote:
Apple is going to use the AMD Fusion platform in a big way
Not exactly an unlikely prediction, unless he's talking about desktop. In which case that's pretty much definitely a no.Quote:
AMD most certainly is going to release a Dual-socket Bulldozer configuration.
I'm assuming he means a server board, which is what Apple utilizes in the Mac Pro (and where he was going by mentioning it in the first place).Quote:
Originally Posted by Apokalipse
13 800 can will be real for 8c Zambezi...But, its not nothing oficialy, we must just waiting :)
That score of 13.88 in C11.5 is 25% higher than the one posted on Donanimhaber. So either the sample for the DH slide had 25% lower clock(not likely) or the score is not real. DH slide suggests that X8 at xxMhz has ~1.9x perf. of 1100T in C11.5,which is around 10.9-11pts. Still a massive leap in performance. 13.89 is therefore hard to swallow(heck even DH slide is very fishy in many regards).
Uhhhhhggggghhhh, these type of posts... who invents what first.It is really irrelevant, companies will design products to sell products, and they make technical decisions separate from one another for different reasons and arrive at different conditions at different times.Quote:
Originally Posted by ohnoitseddy
But to make a point.
Intel implemented an IMC in the 386 days with the 386 SL ( http://fury-tech.com/en/tag/history-...croprocessors/ ) there was nothing really invented here, where you put a memory controller is a design choice of the platform.
The hypertransport bus is based on the EV6, which was DEC, Intel bought out DEC in 1998. Did nothing with the IP, AMD popularized the serial point to point nicely in the K8 line.
3 level cache is nothing new, several non-x86 designs employed a 3 level cache hierachy, the first 3 level cache x86 CPU was produced by Intel, http://www.dailytech.com/16MB+of+L3+...rticle2564.htm didn't help much it was still craptastic netburst.
Core 2 was inpsired by continuing from Dothan, Banias, and Yonah, all those were P6 lineage, I doubt much of the design was inspired by K8... they are two completely different architectures.
Quote:
Originally Posted by JumpingJack
First on-die L3 Perhaps :) Many Processors before utilised off chip L3 caches
Precisely. L3 cache is not an invention either AMD or Intel can lay claim to. But, in fact, Intel did employ a L3 LLC as a first for x86.Quote:
Originally Posted by mAJORD
JumpingJack, I didn't intend to say anything about who invented what first. In fact, I didn't - I used the word "implemented", and with that I did mean in the segment that we were talking about. Please follow the history of this discussion: I was commenting on Informal's notion that Intel was reluctant to implement features implemented by others first. I only wanted to show that in some important cases, this wasn't so. In that, it doesn't matter in the least who invented it first.Quote:
Originally Posted by JumpingJack
informal:Im sorry, my mistake, I thought, its 13 800points in passmark and I watch at your commnet and again at screen and it is 13.8 R11.5 :). So, think, its next fake...Real can be about 10-11p in R11.5
Hmm, im starting to get a bit annoyed about those damn made up scores without proper background... Of course my post is not aimed at you informal.Quote:
Originally Posted by informal
Expectations grow and grow, by that I mean +20%, 35%, + 50% where will it stop? :shrug: Even if Zambezi is competitive with high end core i7 disillusional people will be still expecting a beast with a lower price tag.
Let me tell you now - Not going to happen!
Either AMD has a competitive product that will be priced on par with Intel, which will lead to a potential price war (good for us) or again we will be waiting for a new architecture for them to compete (not good at all).
Athlon vs PIV is very unlikely to happen again looking at the state of both companies R&D resources, that way some people need to wake up and smell the coffee. Overhyping a product never helped anyone... :confused:
Yeah I agree. Therefore I think we should ignore all these leaks until we have one from more respectable source (with at least screenshots of actual hardware).
While I think Bulldozer may have some very nice hardware features I'm not expecting any miracles,not on desktop at least.In servers it will probably rock. In client it just needs to be good enough to be a viable choice Vs six core i7,that's all.
Realistically I'm expecting something like Phenom II vs Core 2 Quad.Quote:
Originally Posted by Shadov
whats LR-DIMM?Quote:
Originally Posted by informal
Charlie wrote an article on this topic some time ago.
sheesh, what a read! :lol:Quote:
Originally Posted by informal
thanks! :up:
http://www.rumorpedia.net/amd-bulldo...d/#comment-525
Compare:Quote:
We have just received a photo of what sender claims to be an engineering sample of AMD Bulldozer (Zambezi) Quad Core processor benchmark score table.
Unfortunately, he/she asked not to publish the original photo (it also has other test score, but for some reason we cant show it (don’t ask why)).
Anyway, here is 3D Mark Vantage CPU score chart (Sandy Bridge vs. Bulldozer)
http://www.rumorpedia.net/wp-content...rbenchmark.png
3.5 GHz Bulldozer matches 4.0 GHz Sandy Bridge.
We don’t really care whether you believe benchmark results or not, as we just publish what we get our hands on. As always, a cat will die for every chart repost in forums, chats, etc.
Intel is screwed up until Ivy Bridge arrives.
As for other score, it’s a power consumption test.
http://scarletwhore.com/wp-content/u...er-CPUmark.jpg
4 BD cores having a 500mhz lead on 4 SB cores/8 threads is kinda unlikely.
well, I am super duper skeptical. but it is not IMPOSSIBLE.Quote:
Originally Posted by undone
it would certainly be good news for the human race if that were the case.
interesting - not long till release? is it q2?
Ugh, I'd rather rumors like these simply not exist. People seem to carry giant grudges when make believe goals aren't reached by actual products. Wishful thinking, I know.
3dmark 06 VANTAGE :confused:
wow...omg this is not even worth a new thread, but it is relevant to amd...all i can say is wtf! dont these ppl have better things to do than make up far fetched stories??? xbit says its dell.....
http://www.bloomberg.com/news/2011-0...-a-target.html
Things are getting more & more interesting...... :shrug:
http://scarletwhore.com/?p=3277
http://scarletwhore.com/wp-content/u...ulldozer21.jpg
This has been already posted and discussed. It's a week old ''news'',most likely fabricated. Check the search.Quote:
Originally Posted by undone
this graphs look s as fake, passmark is unreal and this is too much for 8c...will be about 10points in real think...