Intel says Penryn "complete"

[cwj717]

But high clock speeds are certainly a good thing- probably the best thing for overall CPU performance as I might add- as long as you can keep the instructions per clock (IPC) high and the temperatures down. I think that with the current focus on good energy and thermal characteristics that we'll only see high-clocked CPUs that stay reasonably cool- no more Prescotts.

Exactly. Many AMD fanboys love saying.."clock-speed doesn't matter anymore"....not true. If that's true, then why is AMD pushing K8 to the brink..(@3GHz it's pretty much topped out, with no headroom left). There's even rumors of a 3.2GHz K8...but if GHz doesn't matter, why bother? It's the old double-standard. :? That seems a little far fetched on the 90nm process. Improving their 90nm process at this point in time doesn't seem very likely, at least not much. Maybe with a new rev of the core or some work to their 65nm process?

 

[cwj717]

I was really surprised at the PA-RISC showing on that list. Even some Alphas too 21364? On-die memory controller...

 

[1Tanker]

But high clock speeds are certainly a good thing- probably the best thing for overall CPU performance as I might add- as long as you can keep the instructions per clock (IPC) high and the temperatures down. I think that with the current focus on good energy and thermal characteristics that we'll only see high-clocked CPUs that stay reasonably cool- no more Prescotts.

Exactly. Many AMD fanboys love saying.."clock-speed doesn't matter anymore"....not true. If that's true, then why is AMD pushing K8 to the brink..(@3GHz it's pretty much topped out, with no headroom left). There's even rumors of a 3.2GHz K8...but if GHz doesn't matter, why bother? It's the old double-standard. :? That seems a little far fetched on the 90nm process. Improving their 90nm process at this point in time doesn't seem very likely, at least not much. Maybe with a new rev of the core or some work to their 65nm process?It may have changed now, but there was talk of an FX-76.

edit: Here's a link:
http://www.tgdaily.com/2006/11/03/amd_65nm_roadmap_update/

The lineup consists of the FX-70 (2.6 GHz), FX-72 (2.8 GHz) and FX-74 (3.0 GHz). In the second quarter of 2007, AMD will add the FX-76, which will be the last 90 nm high-end desktop processor with a clock speed of 3.2 GHz.

 

[slim142]

How silly are AMD going to look if Intel launches 45 nm before K8L? :lol:

You will be the silly one becasue a 45nm die shrink won't put a threat to AMD's Barcelona.

Penryn will just be a die shrink of the C2D architechture and some added SSE4 instructions.

You have to remember that AMD will refine 65nm even more and we still don't know Barcelona/K8L (or call it how you want it) final specs.

Also, AMD will debut 45nm with Ultra-Low-K and DC 2.0 in the 2008 timeframe. Definately, intel is not in the lead or the king of the hill like how many tend to believe and not everything is doom and gloom for AMD.

Wow, no wonder why you got 3 votes for 1 star haha
look, 2008 is way too far from now, what about we talk about the present? Intel is leading and kicking ass, and I think they will be for a long time...

 

[cwj717]

But high clock speeds are certainly a good thing- probably the best thing for overall CPU performance as I might add- as long as you can keep the instructions per clock (IPC) high and the temperatures down. I think that with the current focus on good energy and thermal characteristics that we'll only see high-clocked CPUs that stay reasonably cool- no more Prescotts.

Exactly. Many AMD fanboys love saying.."clock-speed doesn't matter anymore"....not true. If that's true, then why is AMD pushing K8 to the brink..(@3GHz it's pretty much topped out, with no headroom left). There's even rumors of a 3.2GHz K8...but if GHz doesn't matter, why bother? It's the old double-standard. :? That seems a little far fetched on the 90nm process. Improving their 90nm process at this point in time doesn't seem very likely, at least not much. Maybe with a new rev of the core or some work to their 65nm process?It may have changed now, but there was talk of an FX-76.

edit: Here's a link:
http://www.tgdaily.com/2006/11/03/amd_65nm_roadmap_update/

The lineup consists of the FX-70 (2.6 GHz), FX-72 (2.8 GHz) and FX-74 (3.0 GHz). In the second quarter of 2007, AMD will add the FX-76, which will be the last 90 nm high-end desktop processor with a clock speed of 3.2 GHz.

Thanks for the link. Thats going to use some power... Whats the TDP for that thing going to be? EDIT: It says 125watts for a DC FX. Ouch.

 

[1Tanker]

But high clock speeds are certainly a good thing- probably the best thing for overall CPU performance as I might add- as long as you can keep the instructions per clock (IPC) high and the temperatures down. I think that with the current focus on good energy and thermal characteristics that we'll only see high-clocked CPUs that stay reasonably cool- no more Prescotts.

Exactly. Many AMD fanboys love saying.."clock-speed doesn't matter anymore"....not true. If that's true, then why is AMD pushing K8 to the brink..(@3GHz it's pretty much topped out, with no headroom left). There's even rumors of a 3.2GHz K8...but if GHz doesn't matter, why bother? It's the old double-standard. :? That seems a little far fetched on the 90nm process. Improving their 90nm process at this point in time doesn't seem very likely, at least not much. Maybe with a new rev of the core or some work to their 65nm process?It may have changed now, but there was talk of an FX-76.

edit: Here's a link:
http://www.tgdaily.com/2006/11/03/amd_65nm_roadmap_update/

The lineup consists of the FX-70 (2.6 GHz), FX-72 (2.8 GHz) and FX-74 (3.0 GHz). In the second quarter of 2007, AMD will add the FX-76, which will be the last 90 nm high-end desktop processor with a clock speed of 3.2 GHz.

Thanks for the link. Thats going to use some power... Whats the TDP for that thing going to be?I have no idea. With the FX-74 using 125w, i'll assume it would be ~135 watts. Judging by the FX7x overclocking ability(3.0-3.1), i would say it'll need a little more voltage...not too mention some very low-yield binning.

 

[epsilon84]

With the FX-74 using 125w, i'll assume it would be ~135 watts. Judging by the FX7x overclocking ability(3.0-3.1), i would say it'll need a little more voltage...not too mention some very low-yield binning.

Talk about beating a dead horse. Seriously, what is the point in releasing a 3.2GHz K8 right before Barcelona? It'll barely match the E6700, and at ~135W as well. Kind of reminds me of Intel's P3 1.13GHz, pushing things until they break...

 

[djplanet]

With the FX-74 using 125w, i'll assume it would be ~135 watts. Judging by the FX7x overclocking ability(3.0-3.1), i would say it'll need a little more voltage...not too mention some very low-yield binning.

Talk about beating a dead horse. Seriously, what is the point in releasing a 3.2GHz K8 right before Barcelona? It'll barely match the E6700, and at ~135W as well. Kind of reminds me of Intel's P3 1.13GHz, pushing things until they break...

Consider that Intel released the Pentium EE 965 shortly before Conroe. There always seems to be a last hurrah of an aging uArch before it's put out to pasture.

 

[DavidC1]

Man. The FUD Barcelona_Xtreme is spitting is beyond normal. I'd even assume he is a disguise of Sharikou. Since he won't understand anyway and there is no point of arguing, I won't respond back to his bullshit.

Penryn
-Code-name for the 45nm mobile CPU based on the Core microarchitecture, which will be the basis of the cores on the desktop/workstation, just like Merom is to Conroe/Woodcrest
-Do we know any other enhancements to Penryn other than SSE4 and 50% more L2 cache?? Honestly we don't, but there are hints there might be enhancements to the FP unit. The possibility is doubled SSE FP Load performance to match K8L. I am 99% sure nobody here knows, but even Dothan had some enhancements over Banias apart from cache
-From the roadmaps, the desktop chips, Wolfdale/Yorkfield won't have a TDP reduction, instead Intel is using the headroom to increase clock speed. Intel states lowering TDP below 65W won't affect much to the marketshare. The primary benefit to Penryn cores is mobile, there will be no TDP reduction with Wolfdale/Yorkfield/Yorkfield XE
-The die size likely will be greater than 120mm2. The reason for that is the actual circuit size reduction for Intel's latest CPU cores have been poor recently. I'd bet that's a combination of power/cost issues. Intel's own 45nm SRAM test vehicle achieved shrink to 60% over the 65nm SRAM. Banias/Dothan/Yonah achieved 70% shrink to both logic and cache(now if some ignorant bunch of peeps don't understand, that means the physical size has reduced to 70%, not die).

 

[DavidC1]

Consider that Intel released the Pentium EE 965 shortly before Conroe. There always seems to be a last hurrah of an aging uArch before it's put out to pasture.

It's different from AMD. AMD's new process isn't better off from the older process because the older process has been optimized beyond sanity. We'll have to wait until AMD's 65nm process has also been optimized silly, then we'll see a benefit over 90nm.

Intel does much bigger and less frequent process changes over time, but they still do. Which is why you see a bigger benefit from Intel than AMD with new processes.

 

[zjohnr]

Evidently the IPs matched up ...

:roll: :roll: :roll: :roll: :roll: :roll: :roll: :roll: :roll: :roll: Isn't that covered in like, well, Internet Jackassing 101 or such? He probably cut class the day they went over it. Arrogance. Turns around and bites you in the arse sooner or later every time.

-john, the essentially clueless redundant legacy dinosaur

 

[jkflipflop98]

The biggest update this time around is the transistors themselves. I guarantee the first thing you'll say when you read the first wolfy benchmark is "Holy shit!"

 

[djplanet]

Consider that Intel released the Pentium EE 965 shortly before Conroe. There always seems to be a last hurrah of an aging uArch before it's put out to pasture.

It's different from AMD. AMD's new process isn't better off from the older process because the older process has been optimized beyond sanity. We'll have to wait until AMD's 65nm process has also been optimized silly, then we'll see a benefit over 90nm.

Intel does much bigger and less frequent process changes over time, but they still do. Which is why you see a bigger benefit from Intel than AMD with new processes.

This trend is beginning to change. Netburst existed in some form from 2000 to 2006, a six year span. K8 has existed since 2003; it probably will be retired this year. That's a four year span. But Intel's roadmap shows a new uArch every two years: Core 2006-2008, Nehalem 2008-2010, and Gesher 2010-?. We've heard K8L things in the AMD corner, but nothing concrete (no actual reviews, anyway). Hopefully K8L will benefit from 65nm and further process shrinks. I'm not sure of AMD's capabilities (sure they're innovative, but do they have the muscle) to develop a new uArch every two years as well.

All the more reason why K8L needs to be damn good and needs to scale pretty far to carry the Green team for probably around three years, like K8 did. Just my opinion.

 

[32nano_meter]

The biggest update this time around is the transistors themselves. I guarantee the first thing you'll say when you read the first wolfy benchmark is "Holy ****!"

cool.

 

[croc]

There are a few of us who'd like to buy you a schooner or two, in a nice quiet pub somewhere just to know what that statement actually means...

 

[jkflipflop98]

As time comes to pass, you will understand. But I'll take the free brew anyways.

 

[croc]

Next time you're in Sydney.... Might even throw in a 'free lunch'.

 

[MU_Engineer]

If Intel ever built a fab in or near Missouri, I think I could find a few buddies who might be persuaded to buy a few pitchers for you in exchange for a tour

 

[enewmen]

Two questions, Jack: Intel supposedly is introducing both MCM (Yorkfield) and monolithic-die quad-core CPUs (Wolfdale) on the 45nm process using the same micro-architecture while an 8-core CPU is not on the roadmap. If the FSB isn't a problem for shuffling data between the two core banks, why go through the expense of making a new die mask and getting the slightly lower yields of chips? Or do you think there will be enhancements like a shared L3 cache for all the cores that requires a monolithic die like the K8L?

The currently slated quad core from what I have read is Yorksfield, some sites have stated it is monolithic, others MCM --- makes not much difference really other than a monolithic of course avoids the snoop contention over the bus. There is onething about the core uArch and the MCM approach --- since cache is shared, from a coherency point of view, there is not much difference in that only two caches must cohere. The roadmap is kinda merky when it comes to Intel's plans past Penryn, we know of Nehalem, and we also hear that Nehalem will integrate the memory controller as well as support CSI, this is likely where the monolithic quad will play in.... for the very reasons you state.

The MCM approach has two distinct advantages and one distinct disadvantage. The disadvantage is already out --- the snoop and coherency across the FSB. The two advantages are of course well known too and you mentioned it -- time to market and yield. I don't think Intel will move to a monolithic core before the end of 2007 on the current architecture, but that remains to be seen. One thing though that there is not clear answer --- will the FSB bottleneck a quad core Penryn derivative. We don't know --- the FBS will jump to 1333 MHz, giving a full 10.5 GB/sec BW to the socket, but the clock speed of the processor will also scale up, reportedly to 3.5-3.7 GHz. 333 is 33% icnrease but so is the rumored release speed (slightly over 33%). So scaling wise, there should be no difference in observed bottlenecks than what we observe for stock Kentsfield on a quad --- which is none.

All in all it will be a wait and see game.... I am working now to find the FSB limitations on the current quad, and once I find that point I can give you a more reasonable estimate. But to answer your questions, the monolithic quad core design is not on the horizon at least not from any clear cut roadmap and 2nd, the next major revision with the IMC/CSI will do away with any debate about FSB issues.

What I wonder is if Intel will keep the large L2 cache with an IMC --- my guess is no, or if they do, there will be a larger gradation of different cache sizes to choose from...

EDIT: If you are looking for what might be exciting or novel from Intel this year....

- Watch 45 nm, the rumor mill is amiss with high-k, if this is true AMD will not hold the performance crown in 2007 and we will need to look to 2008 for what migh be competitive against Nehalem. But with an IMC and the BW that goes with that, well, AMD has their work cut out for them.... it will be interesting.

- I would not be surprised if Intel cuts in a dual FSB on the high end Bearlake chipset and produces a dual socket workstation/DT MB in the same vain as AMD's 4x4 -- just to keep up with the core race as AMD will tout 8 cores on the board when barcelona hits. This is purely speculation... however, they already know how to do it as broadwater already employees a 64 meg Snoop filter chipset with dual FSBs, I could see a DT variety with say a 32 Meg or 16 Meg snoop filter and dual FSBs.

Jack

Any guess if there will be some 1066 fsb yorkfield and if the p965 can run at 1333 (with a 1333 cpu)?

 

[enewmen]

Any guess if the yorkfield CAN run on a 1066 fsb?

I am going to find out --- I am sure it can --- the question is will it run well??

People assumed the FSB would bottle neck a dual core, it didn't, then people though it would bottle neck a core 2 duo and it does not, then people said well no way in hell can a quad core run on a 1067 FSB, and in fact it does very well.... so the ultimate problem we have are people running around saying FSB sucks FSB sucks it bottlenecks the CPU blah blah and they have little or no data showing such a thing exists other than Intel saying it has plenty of bandwidth, which of course no one will believe....

The best example of showing now FSB bottlenecking the CPU was Tom's data for a 2.67 GHz Quad Kents using multipliers to hold the CPU speed and FSB changing from 1067 to 1333, and guess what --- no difference in performance, which means 1067 is not bottlenecking the 4 cores....

This is limited data though so someone needs to prove it..... there is very little data using the right experimental conditions to establish the validity of the statements one way or the other.

Jack

Thanks for checking! I don't care if the yorkfield is a tad slow (with 1066 should do fine like you said). I just don't like to buy a new motherboard every year. "Underclocking" a cpu or "overclocking" a motherboard might be a way to help make mobos and CPUs more compatible? Just a guess.

 

[evilr00t]

Any guess if the yorkfield CAN run on a 1066 fsb?

I am going to find out --- I am sure it can --- the question is will it run well??

People assumed the FSB would bottle neck a dual core, it didn't, then people though it would bottle neck a core 2 duo and it does not, then people said well no way in hell can a quad core run on a 1067 FSB, and in fact it does very well.... so the ultimate problem we have are people running around saying FSB sucks FSB sucks it bottlenecks the CPU blah blah and they have little or no data showing such a thing exists other than Intel saying it has plenty of bandwidth, which of course no one will believe....

The best example of showing now FSB bottlenecking the CPU was Tom's data for a 2.67 GHz Quad Kents using multipliers to hold the CPU speed and FSB changing from 1067 to 1333, and guess what --- no difference in performance, which means 1067 is not bottlenecking the 4 cores....

This is limited data though so someone needs to prove it..... there is very little data using the right experimental conditions to establish the validity of the statements one way or the other.

Jack
Don't... YOU have a QX6700, Jack? Couldn't you test the performance vs. FSB speed yourself?

If you don't have time to do the tests, I wouldn't mind doing them for you :lol:

 

[croc]

Impatiently waiting...

 

[evilr00t]

Yep --- bulding it now....

Got an extra P5W DH a day or so ago, picked up an 850 Watt PSU today and will beging throwing it together tonight.... expect screenshots tomorrow

What I will now have is a multiplier unlocked dual core and a multiplier unlocked quad core.... I already ran the dual core benches both single and multitasking for CPU to FSB speed ratios of 5:1 and no bottlenecking there. I will need to repeat that experiment now on Quad core....

All the screenshots, benchmark, and futurmark links will be provided and a formal writeup is forthcoming.
Jack

Awesome. Data banishes FUD.

 

[enewmen]

But, the QX6700 is not a yorkfield.

 

[enewmen]

It seems you are telling me running a yorkfield at 1066 is an almost non-issue. You are just checking BW.