More about K8L
http://www.overclockers.ru/images/ne.../05/amd_01.jpg
http://www.overclockers.ru/images/ne.../05/amd_02.jpg
Printable View
Ha, wouldn't it suck when we get to 8 cores that one OC'ed badly and the other 7 did like 5GHz :p:
Back in early 2005 AMD claimed that Fab 36 would be 65nm from it's start and begin shipping 65nm cpus in late 2005. In March 2006 the AMD rumor mill claimed that 65nm K8Ls would be shipping in June. K8L launch dates have since been moved back time and time again.
K8L ES chips in December seems a wee bit early for a CPU that isn't scheduled to ship until Q3-2007.
BTW, I enjoyed MAS' joke about Brisbanes overclocking to 3.5-3.6GHz on $80 mobos. Brisbane's high clock speed capability is probably why AMD is releasing them at 2.6G and under while releasing their new flagship FX64, FX70, FX72 and FX74 on 90nm at 125W TDP. Somehow I imagine that this all makes sense to some of you...
If you rely on the rumor mill, you could also believe that K8L has been pulled in from H1 2008 to H1 2007. Don't assign too much credibility to the rumor mill.Quote:
Originally Posted by Fred_Pohl
AMD has been very consistent with its 65nm and K8L schedule over the last year. 65nm products by the end of this year, K8L by mid-2007. There's no real reason to believe either has slipped from these targets.
I dont remember if K8L already has this, but independant frequency and Vcore adjustements are definately in the works according to AMD.Quote:
Originally Posted by Timmay
from what ive read off teh AMD website, "K8L" will have that.Quote:
Originally Posted by v0dka
K8L sounds like its going to be more than just a step or two ahead of Rev.G. is it true that Rev.G processors will have a fourth complex decoder and OOO loads? as you can tell, i dont know anything about processor architecture, but im trying :p
If Rev.G is something special that will give higher performance, thats good since the competetion between AMD and Intel right now is still pretty decent. any performance gains on AMD's part would be good. it definitely sounds like K8L is going to own intel hard.
Mesyn,
I would not underestimate the TV and cellphone market. Not sure if US is heading into recession because I dont have a crystal ball that can tell me what will happen a few months down the road, but semiconductors industry has hit a speed bump already.
Desktop PC will not go away due to the business users, gamers and HTPC crowd. Business PC is about 33% of the overall PC unit volume, and desktop PC is about 55% of the overall PC unit sales. Make your own intepretations. :p
My concern is more on AMD's cashflow constrain because that will seriously hamper their ability to compete against Intel in the long run.
Altair sounds/looks nice, but what's the actual performance increase from to Windsor to Anteres is more interesting. That is a better gauge to the IPC increase of the "K8L" core.
Breakfromyou,
Yes, K8L has increased the width of the instruction fetch/decode stage set to make the pipe wider/smarter at the front. It also adds/improves its EX stage unit (from 64bit FPU to 128bit or something of that sort), so the pipe is also wider at the back. Overall the IPC improvement from K8 is big, but not as big as P4 to C2D (because P4 just sucked too much).
same can be freaking said for the Itanic :stick:Quote:
Originally Posted by LowRun
as for the Risc vs Cisc debate the winner is:
Cisc front end and with a Risc back end. END OF DEBATE
YOU MUST HAVE GOOD SOFTWARE support BEFORE you even release the product. Otherwise all you have is a power piece of silicon that is unused and useless
I think you're getting your rumors and offical news mixed up. AMD claimed 65nm samples in late 2005, which is very different from the process being ready much less having anything available for sale.Quote:
Originally Posted by Fred_Pohl
Also AMD never said that Fab 36 would be 65nm from the start (but they did say it was designed with 65nm production from the start which I think is what got you confused...):
http://www.theinquirer.net/default.aspx?article=26936
"Production at Fab 36 will start at 90 nanometres but during the year will ramp 65 nanometres as fast as it can, said Ruiz. This means that wafer starts, probably for the first four or month, will be created using the 12-inch wafers as a test of concept before the 65 nano tech, bought in from IBM is deployed, just to sort of bed it in."
http://www.anandtech.com/cpuchipsets...oc.aspx?i=2734
"As we mentioned in our initial coverage of Fab 36's grand opening, all parts shipping out of the plant will be 90nm, with AMD converting to 65nm starting in the second half of this year. Fab 36 will be "substantially converted" to 65nm manufacturing by the middle of 2007. Once again, AMD wasn't able to provide us with any sort of concrete details about the ramp to 65nm."
I don't recall any rumors about K8L being out in June at all, and if there were so what? Thats not official news from AMD so it shouldn't matter at all, or do you hold every company responsible for every rumor that isn't true? I did hear that it had taped out in late June 2006, which is something entirly different from a release, and is certainly possible that this is true given the roadmap for K8L that AMD has released (takes about a year from tape out to have chips available for sale, from June 2006 to late Q2/early Q3 2007 would almost be spot on...).
It seems pretty much dead on to me if they've already taped out in June, unless they plan on missing thier set launch date they'd pretty much _have_ to have some Alpha silicon ES's out to mobo OEM's and other hardware vendors for testing around that time and to work out the remaining bugs in the Alpha ES's and build volume for launch.Quote:
Originally Posted by Fred_Pohl
I recall Intel releasing P4's at 3Ghz or so that often OC'd to more than 4Ghz with good air cooling (and they never did release a P4 with a stock clock over 4Ghz did they?), and I also remember many AXP 2500+'s that OC'd well past the top clocked AXP 3200+ as well...Quote:
Originally Posted by Fred_Pohl
OC'ing is a crap shoot and to suggest we'll all be capable of getting early 65nm A64's to 3.5-3.6Ghz on air would be rediculous IMO, but I wouldn't be shocked to see that sort of OC isn't all that uncommon either.
You wouldn't happen to be able to post a link to some numbers would you?Quote:
Originally Posted by vitaminc
Niether do I and my bad for posting a comment suggesting that I did, however the housing market and the credit bubble that has fueled it has been one of the major forces behind the economy these last few years and that all appears to be drying up ATM... What other market segment would you look at to pick up the slack and/or create stimulus in the economy to justify large and continuous increases in spending on luxury goods?Quote:
Originally Posted by vitaminc
So it hits one speed bump and thats it, party over? I do believe that were going to see a slow down in clockspeed and performance enhancements despite Intel's roadmaps, but I don't think things have fundamentally changed YET (a key word in this sentance, notice how large I've made it...), not until they well and truly do run out of process shrinks anyways.Quote:
Originally Posted by vitaminc
I'll interpret this as you agreeing with me then, thanks! :D j/kQuote:
Originally Posted by vitaminc
Understandable, but I think AMD has worked to mitigate this as much as possible by partnering with IBM and others for help with process development.Quote:
Originally Posted by vitaminc
Depends on if the 40% IPC increase over K8 is a conservative or wildly optomistic (or special case only) estimate, frustratingly enough we'll probably have to wait til' benches are "leaked" from the ES's testing in late Dec. or early Jan. or maybe even launch day if AMD decides take the NDA's seriously again.Quote:
Originally Posted by vitaminc
sweet, but i'd like to see some benchies with new amd cpus
I don't believe anything originating from the rumor mill. Particularly when it concerns new AMD product availability dates. Speaking of which, I wonder how RHTT is coming along? ;)Quote:
Originally Posted by oldblue
No confusion here. AMD claimed that they would start producing 65nm chips in H1-2005 at Fab 36 and be in full 65nm production in 2006. Months later, after failing to meet these claims, AMD changed their 65nm claims to those that you provided links to. It's a simple matter of following the timeline. There is what AMD claimed in early 2005, followed by their various revised claims in 2005 and 2006 as their previous 65nm estimates were missed.Quote:
Originally Posted by mesyn191
http://www.xbitlabs.com/news/cpu/dis...408081329.html
“We are qualifying equipment as we speak. Fab 36 was designed to be a 65-nm from the start. We are making great strides with IBM and our SRAM yields are hitting all our milestones,” said AMD’s Automated Precision Manufacturing (APM) director Tom Sonderman in an interview with EETimes web-site.
AMD completed the Fab 36 building in late 2004 and then started installation of the equipment. According to the original plan the company aimed to start qualifications of the factory in the late 1H 2005. AMD Fab 36 is expected to be in volume production in 2006.
“The whole idea is to begin to process 65-nm in the middle of 2005 and bring it into production in 2006. We're certainly on schedule,” Mr. Sonderman added.
Your analogy between immature first revision Brisbane and very mature revision P4/K7 overclocking potential makes no sense at all. Of course there were 3G P4s that OC'd to 4G and XP2500s that OC'd to 2.5G when that was the OC potential of their mature processes and yields. How that relates to Brisbane when AMD's early 65nm yields are limited to 2.6G eludes me.
If the first Brisbanes OC to 3.5-3.6Ghz on $80 mobos with good air-cooling, I'll buy you one. If AMD begins shipping 45nm cpus in mid-2008 as they now claim, I'll buy you one of those as well. Keep in touch. :D
Really?Quote:
Originally Posted by Fred_Pohl
;)Quote:
Originally Posted by Fred_Pohl
is K8L will be built on 65nm?
:o zenith of the california school systemQuote:
Originally Posted by theteamaqua
btw, yes, they are on 65nm.
yesQuote:
Originally Posted by theteamaqua
I don't see how your quote supports what you're saying, it looks like what AMD has been saying all along for the most part...
“We are qualifying equipment as we speak. Fab 36 was designed to be a 65-nm from the start. We are making great strides with IBM and our SRAM yields are hitting all our milestones,” said AMD’s Automated Precision Manufacturing (APM) director Tom Sonderman in an interview with EETimes web-site.
AMD completed the Fab 36 building in late 2004 and then started installation of the equipment. According to the original plan the company aimed to start qualifications of the factory in the late 1H 2005. AMD Fab 36 is expected to be in volume production in 2006.
“The whole idea is to begin to process 65-nm in the middle of 2005 and bring it into production in 2006. We're certainly on schedule,” Mr. Sonderman added.
The only thing that contradicts the roadmap they've got right now would the volume 65nm production in 2006, so it looks like they've slipped a few months somewhere in thier estimates, hardly makes them out to be pathological liars.
I think to expect no improvments in OC yield from early 65nm AMD chips is a little to pessimistic, my post was to show how little stock clocks often matter as far as OC'ing capability was concerned though you do have a point as well.
I never said nor implied that AMD are pathological liars. I was merely pointing out that they typically fail to meet their long-term timeline goals and used Fab 36 as an example. In 2004 AMD said that Fab 36 would be 65nm from the start, open several months before it did and be producing 65nm chips in H1-06. If you go back a little further to 2002, AMD told investors that they would be producing 65nm chips in mid-2005. I'm not saying that it's a BFD but facts are facts and shouldn't be ignored.
http://www.amd.com/us-en/assets/cont...nalyst_Mtg.pdf
As for Brisbane overclocking, I didn't say that they won't overclock. What I said was that expecting 3.5-3.6GHz on $80 mobos was a joke. If early Brisbane yields were great, I don't think AMD would be relying on 90nm process for all of their >2.6GHz cpus.
Unfortunately this is the way the world works sometimes :\. Companies need to make somewhat optimistic predictions in for public relations purposes. Otherwise, they wouldn't get the market share.Quote:
Originally Posted by Fred_Pohl
Not that I agree with it - I think it is dishonest. Perhaps AMD hit some roadbumps along the way. 65nm technology is not easy. 1 nm corresponds to about 4 silicon atom layers, so that transistor gate width is extremely small.
i agree with you , even a big company like intel has hit several road bumpsQuote:
Perhaps AMD hit some roadbumps along the way.
and it won't be the last for either company :)
lets be honest AMD hit alot of bumps on the road to making K8 but they are still around and in the future there will be more bumps but AMD will still be around. What I am really looking forward to is Socket F, Which should not only let you use K8L processors but also UltraSPARC processors, PowerPC processors and just about any other processor architecture out there (should such company wish to use AMD's socket) Not only that but you will be able to use different processors in the same system. you'll be able to have both a PowerPC processor with its VMX unit to do lots of vector math and also have a K8L processor to play Fear. And if you can afford a 4P system, imagine what exotic hardware you can have play together to give you the best performance in the applications that you want as well as the ability to tweak and modify how your OS uses it and also the ability to select a Given core and tell it to be in charge of all your music or whatever. the Possibilities of this is HUGE and I personally can't wait. (on a side note, very excited about FreeBSD's PowerPC port :D )
Semprons give over 50% of AMD's revenue, so i guess (if that unoficial roadmap is true) they're using the 65nm fab fully for their best selling CPU's and not the ones that do not sell very well...Quote:
As for Brisbane overclocking, I didn't say that they won't overclock. What I said was that expecting 3.5-3.6GHz on $80 mobos was a joke. If early Brisbane yields were great, I don't think AMD would be relying on 90nm process for all of their >2.6GHz cpus.
http://www.amd.com/us-en/assets/cont...nalyst_Mtg.pdf
Actually it says nothing of the sort, just that they plan to qualify the 65nm process by 2H 2005, IOW sampling... Looking at slide 4 I could see what you mean, but they give themselves alot of wiggle room and give no firm date for release to market, just a long green bar that is supposed to signify the production phase.
Don't quite a bit of the X2's OC to 2.9-3.1Ghz on air? 3.5-3.6Ghz doesn't seem to be impossible or a joke with a die shrink... Again, what ships as top clocked aint' the same as what is possible with OC'ing, you don't think AMD just runs SuperPi to qualify its chips do you?Quote:
Originally Posted by fred pohl
Does anyone know how much K8L will cost?
actually, contrary to Fred_Pohl, AMD had a quicker time to market on the Opteron 2000/8000 series than Intel did with Woodcrest. by about 3 weeks.
dave
My point in bringing up AMD's overly-optimistic early estimates for Fab 36 and their 65nm shrink timeline was meant simply as a reminder that AMD, like most publically held corporations, tends to provide overly optimistic long term estimates to entice new investors and satisfy existing investors. Right now AMD is saying that they will start shipping 45nm cpus in mid-2008 and I'm weary of seeing some people take such estimates as gospel truth only to forget all about them when that timeline is later amended to H1-09 or H2-09.
As for Brisbane overclocking potential, I'm just saying not to expect 3.5-3.6GHz from the first revision. Later revisions might do that but to expect such impressive overclocks from early revisions that AMD won't validate above 2.6GHz is just setting yourself up for a big disappointment, IMO.
I didn't mean to bash AMD here. I merely meant to temper what I see as unrealistic optimism with a dose of reality.
yea , i know you better then that tooQuote:
I didn't mean to bash AMD here.
things was going good for amd , they did not see the need to go to 65nm , now they do
Yeah, they probably figured to hell with profits, we'll just stick with 90nm for another 6 months while Intel kicks our ass in profit per cpu and performance per watt. :confused:Quote:
Originally Posted by The Ghost
umm what makes you think it is more cost effective to make something on a smaller process? usually inorder to make things on the smaller process you need to spend even more on the wafers which may or may not make the processors cheaper to produce. It all depends on the yeilds. :stick:Quote:
Originally Posted by Fred_Pohl
Quote:
Originally Posted by nn_step
Yields being the same, the smaller process will always be more cost effective. More cpus per wafer equals lower costs and higher profit margins. AMD and Intel don't spend billions on process shrinks and new fabs to lose more money.
and how do we know that the yields are going to be the same ?Quote:
Yields being the same,
it also takes a lot longer to make the wafer , something like 3 months from start to finish
That is the thing when a new process comes out the yields are LESS.Quote:
Originally Posted by Fred_Pohl
infact the manufacturing cost of Early Yonah processors were similiar to the manufacturing cost of DEC's 21064 processors despite the fact that yonah is on 65nm and the 21064 was on a 3 layer 0.75µ process :stick:
Smaller ONLY MEANS CHEAPER IF THE PROCESS YIELDS ARE GOOD.
And just a little fact for you, ALL NEW PROCESSES HAVE BAD YIELDS. Hence AMD is switching to 65nm when it suits them BEST, not before
I thought that was what you were doing too, just didn't agree with all your reasons.Quote:
Originally Posted by Fred_Pohl
:toast:
Huuuhh, they've been out for long enough that you spell them correctly, it's YONAH :stick:Quote:
Originally Posted by nn_step
Bingo, when AMD made the move to 90nm with winchester they released 3000+, 3200+, and 3500+ processors because there was more volume there. While not all of the winchesters clocked great i had a very early 3200+ that hit 2.77ghz stable, (i believe) before the fx-57 even came out. Just because they arent releasing at high clock speeds doesnt mean the process sucks. It makes a lot more sense to have your high volume parts be on your smallest process because you can make a whole lot more of them. Also, since your early process will be spotty with parts that clock and some that clock low you can focus on your mature process for your highest speed low volume parts.Quote:
Originally Posted by alayashu
Now realistically, i think that with the early parts youre gonna see a lot of variation but you will definitely see a good amount hit in the 3.2-3.3 range, not amazing, but it barely happens with current silicon/process.
It's absolutely possible I'm afraid :stick: . I remember numerous S940 Server mobos which was equipped with 2 CPU sockets and 4 DIMM slots laid out only to first CPU. In this situation all memory data requested by CPU2 was redirected to IMC from CPU1 in same way like it's done when you have data in another CPU memory pool (NUMA).Quote:
Originally Posted by informal
But TBH I don't think AMD will made quad core RevG.
Your right, even the 90nm SOI of today makes the 65nm's of intels look bad. Since the 65nm SOI's with Girminanium, Cu, COMS, DSL, and afew other processes alone in normal production will blow intels 65nm processes away. All intel uses is COMS Cu 65nm. Nothing else so its sad and the reason why 90nm AMD's are still better then the 65nm intels.Quote:
Originally Posted by nn_step
Thats why a 90nm 35w amd upto 2.4ghz is much better at performance per watt then conroes upto a E6600. When they come out, because currently it goes upto a 2.2ghz at 35w now, but is still quite impressive consittering its 90nm and not 65nm. Just because the process is smaller doesn't automatticly make it better in anything. Theres more to it then just die size people.
You have to consitter the fab processes themselfs with the arc desine silly.
screw performance per watt, when amd can make processors that can clock to 5.5ghz, please get back to me. this isnt green peace systems, this is xtremesystems... by the way, low leakage transistors which constitute these 35w chips can barely clock over stock, so no thanks, ill take an x6800 that can clock to 4.5ghz on water and it can eat all the power it needs to get to 4.5ghz...
i hope you do , because you are not going to be happy watching the K8L when it comes outQuote:
ill take an x6800 that can clock to 4.5ghz on water and it can eat all the power it needs to get to 4.5ghz...
so then i figure you already tried k8l?
It is mostly because K8L sips at power because it will have a slightly thicker gate oxides, which will slightly reduce switching speed, in exchange for virtually no current leakage and a brand new Record for performance Per watt. they could in theory reduce the gate Oxide thickness in exchange for more speed but that would go against their goal of performance per watt. With the side benefit of making it significantly more reliable than any processor made by Intel.
I honestly wouldn't mind if they made an enthusiast version that ignored heat issues and that just went balls out for speed. I have no problem running large fans or water cooling or for that matter phase change cooling, this is however impracticle so I guess we'll just have to make due with whatever they give us and see how far we can OC it.
Quote:
Originally Posted by nn_step
nn_step and kemist for president!Quote:
Originally Posted by kemist
Are you on drugs or something ? :stick:Quote:
Originally Posted by nn_step
no.. that's what thicker oxide gates do. reduce leakage at the expense of switching speed ;)Quote:
Originally Posted by savantu
Well if you have your electric bill paid for you then that's fine, but i don't, performance per watt FTW! :toast:
I think a certain hardcore fanbois should relax abit. They got the ball and running for an endzone that doesn´t exist. K8L is an EVOLUTIONARY upgrade of K8. The only performance enhancing thing in K8L is absicly its ability for the single cycle SSE like Core 2. However things like macro fusion and such is still out. Speed is also somewhat hyped upwards. Specially when looking on the number we see from AMD. Also K8L gonna fight 45nm Core 2 and a year after Core 3. :fact:
But some L3 cache, single cycle SSE and some improved branches dont give 40%. Nor will the power consumption drop to crazy low levels. Its the same 65nm process, and that same process AMD says 65W for smaller regular K8s. :fact:
This sounds just like Intel fanbois saying Core 2 would be 50% faster overall than K8. Reality gonna be a biatch when it hits :D
So evolutionary, not revolutionary. This aint the K9 or K10. Its a easy hotfix to K8 to make it last a year or 2 longer until they are ready with something vastly better. The K8L is more like the P4D when Intel had to do soemthing about the X2. So the big game aint this turn, but the next one.
For me K8L is revolution in evolution!Quote:
Originally Posted by Shintai
Applying your logic, C2d is a Piii upgrade!!
Go and try to add 1 useful transistor to complex design like K8 or Core then we can debate!
core 2 45nm is just a core size reduction with no improvements , K8L has many improvementsQuote:
Also K8L gonna fight 45nm Core 2 and a year after Core 3.
Doesnt matter much, if a 45nm Core 2 is 3.733Ghz and a 65nm K8L is 3.2Ghz. Aswell as the cost of production. Its factors that cant be left out either. So any clock to clock advantage gets eaten up. And currently Core 2 leads the clock to clock and clockspeed advantage. So process size matters alot.Quote:
Originally Posted by The Ghost
Also 45nm Core 2 does have improvements in form of SSE4 parts and more cache, plus maybe a few minor other parts.
I wanna see K9/K10, K8L is just a bump to try and close the gab abit.
k8l = k9.
you can't judge how it behaves since there are no public benches yet.Quote:
K8L is just a bump to try and close the gab abit.
btw the clockspeed advantage you're reffering to isn't valid as the most powerfull 2 cpus from intel and amd have both 3ghz (woodcrest vs 2220 opty). 65nm vs 90nm.
How much can you OC a K8 today on air? How much can you OC a Core 2 on air? Thats right, about 1Ghz more.Quote:
Originally Posted by alayashu
Why should Intel release a 3.5Ghz Core 2 if they already can clear any competition with a 2.4Ghz. Its about $$$. Whenever AMD comes out with something, you release higher speedgrades and harvest the cash again. Its the same reason we had very expensive X2s until Core 2. Now they are cheap as hell. Its all about competition.
both 3Ghz huh?Quote:
Originally Posted by alayashu
thats as useful comparison as back when Intel P4 reached 2Ghz, and AthlonXP also reached 2Ghz.
how are you getting that the 45nm core 2 is going to be 3.73ghz ?Quote:
Originally Posted by Shintai
i see how you just listed off all the 45nm core improvements , quite impressive , does your core 2 not show sse4 on it ?
remember one thing , this is a first for intel , four cores on one cpu , amd's cpu's was buoilt from the ground up to be multible cores , the conroe was not
no, but if you don't believe me... Feel free to do some reading on Modern fabrication processesQuote:
Originally Posted by savantu
do you remember the pressler ? it ran as high as 4.5ghz on air , why didn't intel release it while it was losing market share against the K8 ?Quote:
Originally Posted by Shintai
right on..:toast:Quote:
Originally Posted by brentpresley
If Intel can continue to keep their pricing/performance like they did with C2D then it will force AMD to do the same; which greatly benefits us, the consumer. Frankly my old days of buying $1K cpu is no more, since C2D clearly demonstrated price/performance cant be touch. In this industry if you choose sides you will lose; I choose the company with the greatest price/performace and if AMD can bring that to the table, then I'm all for it. Very exciting years to come.. :)
Short-term: yes, long-term: noQuote:
Originally Posted by syne_24
cheap chips = less funds for research and development.
Probably he got that from here:Quote:
Originally Posted by The Ghost
http://resources.vr-zone.com//newspi.../Yorkfield.GIF
even though that's about Yorkfield.
I expect K8L to be very strong clock for clock,
if Intel thinks 3.73GHz is needed to counter a 2.9GHz cpu it must be serious.
I'm very curious about how well both K8L and 45nm Core2 will oc.
I hope we'll finally see some real competition, after K8 >> Netburst and Core 2 >> K8.
Do 65nm K8's oc better then the 90nm ones? And how much?
Thermal issues? Core 2 dont have the same issues.Quote:
Originally Posted by The Ghost
No my Core 2 Duo shows SSSE3, or Supplemental SSE3 if you like.Quote:
Originally Posted by The Ghost
AMDs CPUs are build for multiple cores? Thats the dumbest thing I heard in a long time. If you said multiply sockets yes. But the other one is complete lack of the basics. How is it an X2 communicates and make a L1 cache snoop on the other CPU? Yes that correct, more or less like the P4D way without the FSB latency and just remove the MCH with the SRQ. The Core 2 got direct L1 links and shared L2 cache. Why do you think K8L gonna get shared L3? Its to skip another step instead of the SRQ. So thats the most bull thing I heard in a long time. Core 2. L1->L1. K8L L1->L2->L3->L2->L1. And a K8 is L1->L2->SRQ->L2->L1. :fact:
And the Yorkfield will do like Core 2 Duo, just with 4 cores and 1 12MB shared cache. Now, next time. say MULTIPLE SOCKETS. :slapass:
:stick: Yes it will, most of the time to get more speed you need either a better design (not really possible for Conroe without a massive redesign) or more voltage.Quote:
Originally Posted by Shintai
So to give Conroe more speed they would have to increase the (stock) voltage.
And that will start to geometrically increase power usage and Geometrically increase Heat output.
Cept that alot of people can already get 3.3-3.5Ghz on stock voltage. Or like me, 3Ghz on 1.24V :cool:Quote:
Originally Posted by nn_step
same thing , this what core 2 is lackingQuote:
AMDs CPUs are build for multiple cores? Thats the dumbest thing I heard in a long time. If you said multiply sockets yes.
yes each core has a direct link for each core , not every core , core 2 shares L2 cache with all cores , amd cpu's have a direct link thru the x-barQuote:
The Core 2 got direct L1 links and shared L2 cache.
why do you think that amd is going to have such a small L3 cache ? 2megs compared to yorks 12meg cache ? why do you think that amd shrunk the L2 cache for ? because it has L3 cache , shared L3 cache will have little evvect on the K8L other then it will store more info that can be shared with 4 coresQuote:
Why do you think K8L gonna get shared L3?
shows how little you actually know about the K8 , i said it right the first timeQuote:
Now, next time. say MULTIPLE SOCKETS.
i do not see how that means anything for a quad core where there will be more heat in a more confined place , smaller die means that you have less area of contact to get rid of the heatQuote:
Originally Posted by Shintai
Seriously Ghost, get a schematic of both Core 2 and K8 and check. the crossbar is the SRQ interface. There is no direct links in K8 or even K8L. K9 will most likely have direct links like Core 2.
Multiply cores and multiply sockets cant be compared, specially not when you speak about direct links that you didnt even understand in the first place.
Why K8L got small L3 cache? Same reason K8 now got small cache. It hardly got any benefit for K8 unlike for Core 2 with its higher latency to main memory.
Sorry Ghost, but you seem too fanboism or even shilled. No need to fool people or make something up you want to. But trying to twist the facts is not helping AMD in any way. Anyway, i´m done. reality will soon catch up with you, when AMD demonstrates K8L and we see benchmark numbers. If it was a Core 2 killer, you would have seen the numbers by now to stab Intels Core 2 sales where it hurt.
funny I didn't see Any K7 or K8 benchmarks 9 months before release. :stick:Quote:
Originally Posted by Shintai
Ghost you really have no jack idea on what you are talking about.Quote:
Originally Posted by The Ghost
There's a extremely big difference between multi socket and multi core designs, and K8 isn't really designed for multi-core otherwise you would seen quad/hex/oct K8 already.
K8 is better at multi socket design and is jack :banana::banana::banana::banana: for multi core design. Exactly why AMD need to make K8L for the shared L3 cache and core to core arbitration.
If people can get 3.3-3.5GHz on stock, that means Intel could release higher frequency chips in the same TDP rating. Not magic.
maybe you should read more posts here instead of posting , unless it is something that you do knowQuote:
Originally Posted by vitaminc
http://www.xtremesystems.org/forums/...t=55889&page=3
this is obviously some thing that you do not know about
we know who knows what and who does not know , i said that amd designed the cpu to be dual core from the start , would you believe it if amd said it ?Quote:
Originally Posted by Shintai
http://multicore.amd.com/en/Products/
Dual-Core Processor Overview
AMD64 processors were designed from the start to add a second core
Port already existed on crossbar/SRI
keep on going , you are showing exactly what you do know
or if you could please use some primary sources as your backup. both AMD and Intel publish quite a few white papers and specs, and I am sure an AMD enthusaist like you should have read many.Quote:
Originally Posted by The Ghost
and no, multi core and multi socket cpu design are NOT the same thing.
K8L looks cool, but I think it will have a serious clocking problem. K8 already has a clocking problem compared to Conroe and I can't see K8L clocking higher than K8. I think they might even have to drop the clock a little for the highest-end part to about 2.6 or 2.8 GHz max. And this biatch will run hot IMO.
So at stock it might win some from Kentsfield, but it won't overclock worth a :banana::banana::banana::banana:. That's right, I am betting Kentsfield will rape K8L while overclocked just because it clocks so well.
Not everyone needs a Quad Core. I, personally, don't really. What about dual core? If K8L will be strictly a Quad Core part, dual core market (majority of perofmance market anyway) will be owned by Conroe. I don't think anyone wants to argue how hard Conroe kicks K8's ass on any tech process.
please god let this thing rape C2D...
With twice the # of cores, it better 'rape' C2D.Quote:
Originally Posted by GOATSLAYER
do you need to be spoon fed ? check the link right above your post , that is a amd sourceQuote:
Originally Posted by vitaminc
try doing a little more reading instead of insisting that you know all about amd cpu's , the best way to defeat them , is to read all about them
Even with the same amount of cores, it better rape C2DQuote:
Originally Posted by vitaminc
Remember Dothan vs K8? I think that will kind of be Kentsfield vs K8L. Higher clocks at stock but Kentsfield should win because of great overclocking. The clock delta will be larger.Quote:
Originally Posted by Daved+
4 GHz Kentsfield on air is very possible. K8L? Don't make me laugh ;)
Seriously ghost, this is your so called "direct link". Give us a break by now.Quote:
Originally Posted by The Ghost
http://www.penstarsys.com/reviews/cp...0_939/xbar.jpg
Maybe that can help you understand it, since you obviously got huge troubles with it. Now go study or back down on the ultra fanboism. You are making things up that aint there.
Core 2 is currently 20% faster or so than K8. We still haven´t seen numbers of K8L. So if a K8L can afford to lag behind at around 800-1000Mhz or 35% or so. We talk about advancement in design at over 50%. Sorry...its not that good :mad:Quote:
Originally Posted by LOE
And it would be the biggest breakthrough in 10 years or more. It just screams unrealistic when its improvemnts of an existing arch.
To compare, Core 2 is around 15-20% over Core. However when all the "mobile" restrictions got removed it could reach higher in clock. With K8->K8L you aint turning a mobile CPU into a desktop. But desktop to desktop or server to server.
Also on a sitenode, all info sofar points to a 2.7-2.9Ghz K8L quadcore at launch.
if AMD launches a 2.9 cpu to counter an 3.73 cpu (again these freq are fictive more or less) then this pretty much says it all about their new core. But, again, we dont have any benches of Altair/Yorkfield to judge.
you say it as a rule, that a cpu must come from an mobile design to be good, wich equals stupid logic, no offence.Quote:
With K8->K8L you aint turning a mobile CPU into a desktop. But desktop to desktop or server to server.
No i dont, but the Core design was hold back by it. Just compare Conroe and Merom today. So its not some rule, it just had a higher margin of boost, since you could change the power envelope on a design from 31W to 65 and 75W. Its no secret that a Core Duo without restraints could easily follow a X2.Quote:
Originally Posted by alayashu
And where is AMD launching a 2.9Ghz core against a 3.73Ghz? No thats something you make up. The performance crown is a nice thing, but AMD dont earn its money on that. They earn their money on cheap CPUs and the socket scaling of the Opteron. What AMD is releasing is a K8L server design that Intel cant match yet in 4sockets and above. Thats money! :fact:
Look on AMDs roadmap, if AMD needed to take back the performance crown I am sure we would see 65nm FX chips as the first thing. But we dont even se 65nm FX chips before atleast Q3 2007. Its about volume and money, not some epeen contest :slapass:
Today not even 20% of AMD sales sofar I recall is dualcore. And for Intel its also not much higher if not smaller.
Overhyping a product wont make it any better. And you gonna get a reality check real soon when AMD releases the benchmark numbers. What do you expect, K8L that is a K8 with single cycle SSE and some improved buffers and prefetch to beat the K8 with over 50% clock for clock? Single cycle SSE vs 2-3 cycle and all the fusion macro, better prefetch, 4 issue wide and the memory "guessing" system improved Core 2 around those 15-20% over Core. So dont tell me K8L gonna make a 50%+ boost on less improvements.
Else we can just sit and write fiction aswell how K9 gonna beat K8L with 6000% and use 0.0001W. Lets try stay with some facts atleast.
Sorry for the rant, but all this fiction gearing up for new core archs lately is redicilous. It was the same thing about Core 2 with 50% ending as 15-20% and 3.33ghz ending on 2.93. People tend to believe anything like it was some football match cheering on their team.
If AMD feels confident enough about the performance and design of this next chip, they should use the K9 name. :)
sorry , your diagram does not represent the order on which things are done on the cpuQuote:
Originally Posted by Shintai
amd's dual core and quad core talk to each and every core , at the spped of the core , something that conroe based cpu can so
next time try and get some pictures that represent what we are tallking about , not out of comic books
http://img8.exs.cx/img8/831/origarch...20017pc.th.png
and here is from 2001, showing they already had the links in for dual-core ;)
http://img8.exs.cx/img8/4409/orignb20010ni.th.png
and just imagine how much you read , i had already posted a link to these pictures
some more spoon feeding
You're terrible missinformed my friend.For Intel the figure is 60+ % right now and will end 2006 with 90% which is a little bit more than AMDs' 25%.OTOH AMD can flood the market with cheap *ss Sem:banana::banana::banana::banana:s and retain or even gain marketshare ( see Dell ).Quote:
Originally Posted by Shintai
That my friend was because of AMDs' innability to compete.Intel expected AMD to ramp faster both on DC and 65nm.When they saw the reality they downgraded Conroe XE from the original 3.33GHz to a mere 2.93GHz.At the same time power dropped from the original 95w to 75w.Quote:
Sorry for the rant, but all this fiction gearing up for new core archs lately is redicilous. It was the same thing about Core 2 with 50% ending as 15-20% and 3.33ghz ending on 2.93. People tend to believe anything like it was some football match cheering on their team.
It wasn't " we can't do a 3.33GHz Conroe" but it was more like " we don't need a 3.33GHz Conroe ".
this interesting , just where did you get this bit of information from ?Quote:
Look on AMDs roadmap, if AMD needed to take back the performance crown I am sure we would see 65nm FX chips as the first thing. But we dont even se 65nm FX chips before atleast Q3 2007. Its about volume and money, not some epeen contest
Today not even 20% of AMD sales sofar I recall is dualcore. And for Intel its also not much higher if not smaller.
amd passed the dual core being 20% of it's sales for a long time , so has intel
again, multi core design are completely different from multi socket design. you are an idiot if you believe those 2 are the same thing.Quote:
Originally Posted by The Ghost
Both K8 and K8L's multicore design are not as tightly integrated as C2D, but a lot more scalable because its designed for multi socket systems.
BTW, Shintai's picture is directly out of AMD's presentations, if that's your idea of comic books.
I agree with the consensus above. With 2 more cores, K8L better be able to beat the Core2 Duo by a significant margin.
Hopefully AMD has something else in the works to succeed the K8 family.
K8 was made from the start to be a server cpu , so there fore it was designed to work in a multi-cpu systemQuote:
again, multi core design are completely different from multi socket design. you are an idiot if you believe those 2 are the same thing.
Both K8 and K8L's multicore design are not as tightly integrated as C2D, but a lot more scalable because its designed for multi socket systems.
amd was designed from the start to be a multicore cpu , i provided the link for it above , this is what you said above
yes there is a difference between muli-core and multi-socket support , but amd was designed from the start to do bothQuote:
Originally Posted by vitaminc
here is my link and quote again , please note the source , if you know more then the source knows , then i would like to know about it and how you came to that information
http://multicore.amd.com/en/Products/
Dual-Core Processor Overview
AMD64 processors were designed from the start to add a second core
Port already existed on crossbar/SRI
do you need a link to prove that opteron was designed from the beginning to be a server cpu ? that should not be too hard to find
and what in the world makes you think that conroe based cpu's are more tightly intergrated then and's K8 ??
i know where it comes from , but it does not accurately put the crossbar in the right place , the cross bar operates at the same speed as the cpu and the L1 cache , L2 cache is a lot slowerQuote:
BTW, Shintai's picture is directly out of AMD's presentations, if that's your idea of comic books.
1. If you know where Shintai's lil pix came from, you should not put it down that harsh with comments like "comic books" and such. Blame AMD's marketing people perhaps, instead of shooting the messenger.Quote:
Originally Posted by The Ghost
2. Now you are acknowledge there is a difference between multi core and multi socket. And may I kindly refer you to Intel's website for their white papers on CD/C2D architecture. They have detailed information on how much better the dual core design is better than K8. A pro techie like you must know about Intel's architecture inside out, right?
ENOUGHQuote:
Originally Posted by vitaminc
the difference between multi-core and multi socket is this:
Multi-core:Multiple Cores in ONE SOCKET
Multi-Socket: Multiple Sockets on ONE MOTHERBOARD
K8 DUAL CORE IS MULTI-CORE
AS IS PENTIUM D's MCM PACKAGE
STOP THE :banana::banana::banana::banana:ING BICKERING YOU :banana::banana::banana::banana:ING FANBOYS :slapass:
1. You are oversimplifing all that XBAR/SRQ/Shared L3/Arbitor/Shared L2/fan out/etc in multi core design.Quote:
Originally Posted by nn_step
2. You are also ignoring cooling/substrate design/mechanical design/PCB layout for multi socket design.
3. I use both AMD/Intel systems whereever they excel.
because it did not give a realistic view of where the crossbar is located , and that where his pic showed it would make some one think that it does not communicate with other cores on the same socketQuote:
1. If you know where Shintai's lil pix came from, you should not put it down that harsh with comments like "comic books" and such. Blame AMD's marketing people perhaps, instead of shooting the messenger.
dual cores talk to each other at the speed of the cpu , not the speed of the L2 cache , not at the speed of the fsb
that is where conroe is lacking
yes right , you think that i'm going to waste my time looking for something that does not exist ?Quote:
2. Now you are acknowledge there is a difference between multi core and multi socket. And may I kindly refer you to Intel's website for their white papers on CD/C2D architecture. They have detailed information on how much better the dual core design is better than K8. A pro techie like you must know about Intel's architecture inside out, right?
if you really think that it exists , you go find it , this will give everyone a nice break while you try to do that
1) I am following the definition http://en.wikipedia.org/wiki/Multi-coreQuote:
Originally Posted by vitaminc
2) What does cooling have to do with A name :stick:
3) Then stop bickering like a :banana::banana::banana::banana::banana: :slap:
Just stop debating, :slap:Quote:
Originally Posted by The Ghost
You are making an ASS out of yourself.
You two are Both right and Both very wrong.
Argueing on the internet is like running in the special olympics
Even if you win, you still are retarded
the rest of your post has been dealt with already so lets go to this part of your post
are you trying to compare a amd 90nm cpu with a intel 65nm cpu ?Quote:
2. You are also ignoring cooling/substrate design/mechanical design/PCB layout for multi socket design.
like has been proven the K8 has been designed from the concept to be dual core cpu , intels design to match amds for being dual core design , does not even come close to amd's design
at best each one of intels cores can only talk to the other at the bandwidth speed of the L2 cach on intels cpu , and for intels quad core to do that it can only talk to the other half of the cpu at fsb speed
amd's 65nm cpu will be out soon enough to let us know just how cool these cpu's will be
1. there are multiple arbitors between cores for conroe, to handle cache/if/id disputes, which is similar to xbar/srq except it's like glue between 2 cores thus better dual core design. But xbar/srq are designed for multi processor/socket platforms, so they scale better.Quote:
Originally Posted by The Ghost
2. cooling requirements as in the # of cores you can put in one socket without heating up like GPU, substrate design as the # of dies you can put on one substrate with a decent yield, etc. there's a very good reason why people do 2/4/8-way motherboards instead of just using dual/quad/oct cores.
I'd like to see some strong competition from AMD to counter Conroe.
I second this... I want AMD to hit them high, and hit the low...
~Mike
And you, in your infinite wisdom , consider this to be a disadvantage compared to a crossbar ? :gay:Quote:
Originally Posted by The Ghost
Let me ask you : do you have any clue whatsoever on the BW and latency of the shared L2 vs. crossbar ?
Well, without any competition from AMD, it could be a dull point in CPUs. Hope they release it soon enough. Q3 of 2007. There goes my upgrade plan for next year :(
i have a question is really that it is K8L because is AMD processor 4 x 4 but i no talk that is was K8L