Will the Presler Pentium Extreme Edition lay the Smack down a FX-60?

[Hard Ball]

Originally posted by: PC Surgeon
Intel Pentium Extreme Edition 3.6 Ghz (Presler) LGA 775 w/Hyper Threading 1066FSB Dual Core 2 x 2mb Cache

vs.

AMD Athlon 64 FX60 2.6 Ghz Toledo 1GHz FSB 2 x 1MB L2 Cache Socket 939 Dual Core Processor

Which will perform beter?

Pentium XE 955 would not really have a prayer against the FX-60, it still has a lot of trouble on most benchmarks against the X2 4800+:
http://reviews.cnet.com/Intel_Pentium_E...55/4505-3086_7-31637916-2.html?tag=nav

Conroe, on the other hand, at 3.33GHz, would beat the FX-60 into mushed pulp, if it ever comes out.

 

[dmens]

Originally posted by: Vee
According to Ruiz "underpromise" and "overperform" is AMD policy, so we're not supposed to hear much. In the long run it may prove to be an effective counter to Intel's extremely early paper launches which apparently are aimed at having people waiting for a year or more, rather than buying AMD.

Don't write off K8 yet. I think it will both clock higher and reach higher ipc.
Also, the Conroe's desperate FSB speed and large cache size is an admission that it is memory bottlenecked. Allendales cut down cache is admission that it will be expensive.

AMD focus is on multicore/multicpu scaling. New hypertransport, direct connection architectures and level 3 cache are in line. They have also expressed the intention to launch a quadcore cpu at the "mainstream desktop" in 2007. But my guess is that it'll still be K8 derived.

...ummh, and maybe it's architectures.

Which puts a spin on a confused issue. Maybe it's not K8L or K9 or K10. Maybe it's and. One modified core and then two completely new ones. One lean core and one muscle core. Just another example of doing the multicore homework.

(It doesn't matter what they're called. For the scope of this post, K8L, K9 and K10 are adequate identifiers.)

Wouldn't a K8 quadcore be a bit toasty? Considering its current power draw, even with a switch to 65nm, to match merom's performance @ 3ghz using a K8, my estimate those cores would have to be running at 3.5-3.6ghz. imo if AMD is coming out with a quad in 2007, it'll be "K10", and its something heavily reworked, even if it still retains similarities to K8. Well, even merom retains similarities to P6, heh.

as for "skewed" cores, is that really a good idea given the huge hike in design and validation efforts required for such a move? moreover, as cell demonstrated, such an approach requires software synchronization. of course, cell is heavily skewed, to the point where ps3 devs refuse to make games for it. but even with a "soft skew" as suggested, the amount of software support required is quite daunting and probably not worth the risk.

In regards to merom's large L2 as proof of its bandwidth limitations, since woodcrest will be using FB-DIMM, is it still an issue? probably not. with main memories and worksets getting larger, last level caches must grow regardless.

 

[Fox5]

Originally posted by: Vee

Originally posted by: TSS
...preslers are far from interesting. Conroe/Merom will be the first new intel procs that will be. but personally, im more looking forward to K10. i've heard enough about conroe beeing based on the pentium M/3, and so on and so on... but i have yet to hear anything about the new AMD architecture (which we'll see in what? 2007?). so thats where my attention is going to be

According to Ruiz "underpromise" and "overperform" is AMD policy, so we're not supposed to hear much. In the long run it may prove to be an effective counter to Intel's extremely early paper launches which apparently are aimed at having people waiting for a year or more, rather than buying AMD.

Don't write off K8 yet. I think it will both clock higher and reach higher ipc.
Also, the Conroe's desperate FSB speed and large cache size is an admission that it is memory bottlenecked. Allendales cut down cache is admission that it will be expensive.

AMD focus is on multicore/multicpu scaling. New hypertransport, direct connection architectures and level 3 cache are in line. They have also expressed the intention to launch a quadcore cpu at the "mainstream desktop" in 2007. But my guess is that it'll still be K8 derived.

...ummh, and maybe it's architectures.

Which puts a spin on a confused issue. Maybe it's not K8L or K9 or K10. Maybe it's and. One modified core and then two completely new ones. One lean core and one muscle core. Just another example of doing the multicore homework.

(It doesn't matter what they're called. For the scope of this post, K8L, K9 and K10 are adequate identifiers.)

Do you mean an asymmetric multi-core, or several different processor designs? AMD has pretty much always stayed to one processor design across all its products, only recently with the Athlon 64s/FX/Opterons/Turions are you seeing slight differences between the processors. Intel's gone a bit further with differentiation than AMD has.

Also, from what I recall, K9 was just going to be a quadcore K8L.

 

[Viditor]

Originally posted by: dmens

Originally posted by: Vee
According to Ruiz "underpromise" and "overperform" is AMD policy, so we're not supposed to hear much. In the long run it may prove to be an effective counter to Intel's extremely early paper launches which apparently are aimed at having people waiting for a year or more, rather than buying AMD.

Don't write off K8 yet. I think it will both clock higher and reach higher ipc.
Also, the Conroe's desperate FSB speed and large cache size is an admission that it is memory bottlenecked. Allendales cut down cache is admission that it will be expensive.

AMD focus is on multicore/multicpu scaling. New hypertransport, direct connection architectures and level 3 cache are in line. They have also expressed the intention to launch a quadcore cpu at the "mainstream desktop" in 2007. But my guess is that it'll still be K8 derived.

...ummh, and maybe it's architectures.

Which puts a spin on a confused issue. Maybe it's not K8L or K9 or K10. Maybe it's and. One modified core and then two completely new ones. One lean core and one muscle core. Just another example of doing the multicore homework.

(It doesn't matter what they're called. For the scope of this post, K8L, K9 and K10 are adequate identifiers.)

Wouldn't a K8 quadcore be a bit toasty? Considering its current power draw, even with a switch to 65nm, to match merom's performance @ 3ghz using a K8, my estimate those cores would have to be running at 3.5-3.6ghz. imo if AMD is coming out with a quad in 2007, it'll be "K10", and its something heavily reworked, even if it still retains similarities to K8. Well, even merom retains similarities to P6, heh.

as for "skewed" cores, is that really a good idea given the huge hike in design and validation efforts required for such a move? moreover, as cell demonstrated, such an approach requires software synchronization. of course, cell is heavily skewed, to the point where ps3 devs refuse to make games for it. but even with a "soft skew" as suggested, the amount of software support required is quite daunting and probably not worth the risk.

In regards to merom's large L2 as proof of its bandwidth limitations, since woodcrest will be using FB-DIMM, is it still an issue? probably not. with main memories and worksets getting larger, last level caches must grow regardless.

I think Vee is spot on here...Ruiz is practically shouting that AMD is sandbagging.
1. As to quad core temps, don't forget the new SiGe strained silicon process. That should reduce power/performance tremendously

2. Merom looks good on paper, but there are some serious hurdles left for us to see...like can it come close to fully utilizing it's 4-issue core (shorter and wider MIGHT be better, but it's by no means a slam dunk)? Or, as Vee points out, how big will the FSB bottleneck be with so much data? I don't think we are at the point where we can compare it today's processors, let alone what will be available when it's released.

3. With AMD licensing the new Zram, we could be seeing much larger caches (my guess is that we will see L3 cache) at lower power utilizations. In addition, in this article Phil Hester is talking about adding attached copressors using coherent HT (cHT) links. This is very cool stuff!

4. As for skewed cores, Hester may have been touching on that when he said:
"If you just think in traditional ways about every core running at the same clock rate, I don't think you will see a lot of performance improvement. If you had a way to vary the voltages and clock rates of individual cores, you could see some pretty nice performance improvements"

 

[dmens]

1. Even if you buy their "40% lower power" number which is pretty insane to begin with, that would still not be enough to get a K8 quad part that is within power boundaries when clocked up to match perf in 2007.
2. Again, will there even be a bottleneck with a FB-DIMM setup at that speed? As for 4-wide issue, that is talking about instruction fetch, which is never a problem if you have a half decent predictor.
3. Yes that is very interesting. I would like to see power numbers on such a cache.
4. No that is different. Vee and I were referring to cores with different uarch on the same die, which is far more complicated.

 

[irwincur]

God, there are some idiots in this forum. Intel will be behind for two more quarters - they have already admitted that. Even then, the professional assumption is that AMD will be standing still. Yes, there will be hints of light coming from Intel near mid Summer but AMD will probably hold the lead into next Winter.

 

[openwheelformula1]

Proof: http://www.gamepc.com/labs/print_conten...4523665894662F4F68F4FF079A1E652796B37B

Until Conroe/Merom comes out, Intel has to give away the performance crown to AMD. Althogh AMD doesn't have the bang for the buck like it used to w/ the XPs, you still get a better performer for similar price. The only problem for AMD in my eyes, is that they don't have a super low end to compete w/ 920. X2 3800 performs better than 930 and most of the times better than 940. At this very moment, AMD is simply a no brainer. Intel has closed the gap w/ 65nm process, but not close enough imo. AMD's 65nm is just around the corner as well. I am very interested to see Intel's Conroe. I am gettting sick of building AMDs now, come on Intel give me something good in H2 2006. I haven't truely want to touch a Intel since my last Northwood 2.4c.

 

[Jeff7181]

Originally posted by: PC Surgeon
Its is 2.6 for the FX60. My mistake. But I thought that Intel having dual core, 65nm process and Hyper threading would have made it a better CPU since it is another generation beyond AMD.

See... Intel's markering DOES work. They have people believing the Prescott is a generation ahead of the Athlon 64. Wake up AMD... you could spread some FUD yourselves and sell more processors!

 

[Markfw]

The only problem for AMD in my eyes, is that they don't have a super low end to compete w/ 920.

Well, not for the processor, but when you need a $200 mobo to get to CLOSE to the same performance levels, the 3800 ends up cheaper and still faster. Even if you could get a $100 mobo that would OC to 4 ghz, the total package is within $20, and the 3800 still wins in performance.

 

[Vee]

Originally posted by: dmens
4. No that is different. Vee and I were referring to cores with different uarch on the same die, which is far more complicated.

No. You ran away with yourself. That's not what I was suggesting. I was suggesting that AMD, instead of, like Intel, basing all their CPUs on the same common core, maybe will base them on two different cores. Thus single and dual core Meroms and Allendales will face dual-, triple and quadcore K9s. Woodcrest and Itanium will face K10.

(It doesn't matter what they're called. For the scope of this post, K8L, K9 and K10 are adequate identifiers.)

 

[NoStateofMind]

Originally posted by: Jeff7181

Originally posted by: PC Surgeon
Its is 2.6 for the FX60. My mistake. But I thought that Intel having dual core, 65nm process and Hyper threading would have made it a better CPU since it is another generation beyond AMD.

See... Intel's markering DOES work. They have people believing the Prescott is a generation ahead of the Athlon 64. Wake up AMD... you could spread some FUD yourselves and sell more processors!

You are exactly right. Those that don't ask (glad I did) would be suckered in to thinking 2 cores plus two virtual cores would kill a regular dual core AMD. I agree, AMD should capitalize it's proffits now while they have the performance crown. That is concidering what others have posted regarding AMD possibly losing it's crown by the end of '06.

 

[TraumaRN]

Lets take a moment to pause and enjoy this 'review' posted on Newegg....

AMD user's fear this... (yeah I?m talking to you FX-60 users too!!!)

Pros: This is the first CPU made at 65nm process, cuz of this it will overclock to speeds normally not possible with any other CPU, and it does it at lower volts as well. With just a heat sink and fan it w...
ill reach 4.3GHz, water cooling around 4.4GHz - 4.7GHz, and with a VapoChill LS it has easily reached 5.4GHz - 5.7GHz and should POST at 6GHz even with no motherboard mods yet. That's two cores at that high of a stable overclock then with Hyper Threading on top of it just makes this chip undefeatable...
If you wanna go really extreme and get colder then what the VapoChill LS can provide the possibilities are limitless. While AMD user's I can't say the same, as they have the "cold bug" issue when put under extreme cold (sometimes with just the VapoChill LS) which causes the system to crash. I don't know about you, but I don't like limits on my computers...

Here's one of the many reviews about this CPU.

Cons: Sorry nothiing bad about this amazing processor. Ok so maybe the price lol.... More »


Other Thoughts: Get one while you still can they are going fast!

Somehow sounds like an insane fanboi or intel employee spreading FUD...yay for reviews at Newegg :roll:

 

[Markfw]

Sounds like fatty !

 

[stevty2889]

Originally posted by: DeathBUA
Lets take a moment to pause and enjoy this 'review' posted on Newegg....

AMD user's fear this... (yeah I?m talking to you FX-60 users too!!!)

Pros: This is the first CPU made at 65nm process, cuz of this it will overclock to speeds normally not possible with any other CPU, and it does it at lower volts as well. With just a heat sink and fan it w...
ill reach 4.3GHz, water cooling around 4.4GHz - 4.7GHz, and with a VapoChill LS it has easily reached 5.4GHz - 5.7GHz and should POST at 6GHz even with no motherboard mods yet. That's two cores at that high of a stable overclock then with Hyper Threading on top of it just makes this chip undefeatable...
If you wanna go really extreme and get colder then what the VapoChill LS can provide the possibilities are limitless. While AMD user's I can't say the same, as they have the "cold bug" issue when put under extreme cold (sometimes with just the VapoChill LS) which causes the system to crash. I don't know about you, but I don't like limits on my computers...

Here's one of the many reviews about this CPU.

Cons: Sorry nothiing bad about this amazing processor. Ok so maybe the price lol.... More »


Other Thoughts: Get one while you still can they are going fast!

Somehow sounds like an insane fanboi or intel employee spreading FUD...yay for reviews at Newegg :roll:

Hey now, none of the guys I work with go spreading fud like that, don't go insulting Intel employees..

 

[dmens]

Originally posted by: Vee
No. You ran away with yourself. That's not what I was suggesting. I was suggesting that AMD, instead of, like Intel, basing all their CPUs on the same common core, maybe will base them on two different cores. Thus single and dual core Meroms and Allendales will face dual-, triple and quadcore K9s. Woodcrest and Itanium will face K10.

(It doesn't matter what they're called. For the scope of this post, K8L, K9 and K10 are adequate identifiers.)

Sorry, I see what you're saying now. So is there a K9 core? If it is K8 based, it will probably run into the thermal issues I raised.

 

[Cooler]

Originally posted by: dmens

Originally posted by: Vee
No. You ran away with yourself. That's not what I was suggesting. I was suggesting that AMD, instead of, like Intel, basing all their CPUs on the same common core, maybe will base them on two different cores. Thus single and dual core Meroms and Allendales will face dual-, triple and quadcore K9s. Woodcrest and Itanium will face K10.

(It doesn't matter what they're called. For the scope of this post, K8L, K9 and K10 are adequate identifiers.)

Sorry, I see what you're saying now. So is there a K9 core? If it is K8 based, it will probably run into the thermal issues I raised.

i heard they are going to K10 .

 

[Viditor]

Originally posted by: dmens

Originally posted by: Vee
No. You ran away with yourself. That's not what I was suggesting. I was suggesting that AMD, instead of, like Intel, basing all their CPUs on the same common core, maybe will base them on two different cores. Thus single and dual core Meroms and Allendales will face dual-, triple and quadcore K9s. Woodcrest and Itanium will face K10.

(It doesn't matter what they're called. For the scope of this post, K8L, K9 and K10 are adequate identifiers.)

Sorry, I see what you're saying now. So is there a K9 core? If it is K8 based, it will probably run into the thermal issues I raised.

Hester mentions what Vee is talking about in the article as well...

We are evolving to what I'd say are a minimum of two brand-new core design points, new microarchitectures from the ground up. One is aimed at mobile computers and the very low-power space. Another is optimized for the high-end server space. The question we have now is: Can you pull down the server space, or pull up the mobile, enough to cover the desktop?

Phil Hester interview

 

[ahock]

Presler EE can beat AMD FX-60 when it is overclock to 4.26GHz. This is base from several benchmarks already released. Good thing for Intel is that AMD FX-60 is not a good overclock type processor. You can go beyond 4.26GHz with good cooling system. This is the reason why Dell is going to release their XPS renegde.

Somehow Presler is Intel's answer to AMD in enthusiast realms. The only trade off is heat but peformance wise based from benchmarks is beats AMD.

 

[dmens]

I saw a 5ghz presler today on air. It was using a badaxe mobo that can supply 210A to the CPU. I'm not kidding. The fans were loud like mowers.

Seems like the main reason preslers in public are not hitting freqs the internal labs are seeing is because the power regulators on the mobos are not sufficient.

In regards to the two core design, one can say intel is doing some similar with xscale + core, although we already know core will cover desktop and large portions of the mobile market as well.

 

[Cooler]

Originally posted by: dmens
I saw a 5ghz presler today on air. It was using a badaxe mobo that can supply 210A to the CPU. I'm not kidding. The fans were loud like mowers.

Seems like the main reason preslers in public are not hitting freqs the internal labs are seeing is because the power regulators on the mobos are not sufficient.

In regards to the two core design, one can say intel is doing some similar with xscale + core, although we already know core will cover desktop and large portions of the mobile market as well.

hmm i wonder how long till they fix these for the public 5ghz is very good for oc on air. also they will be releasing a 3.6 GHZ pressler in april.

 

[Markfw]

Why would somebody want a processor that takes that much to cool, or sucks that much power ?

I guess fanboys never die.......

 

[TuxDave]

Originally posted by: Markfw900
Why would somebody want a processor that takes that much to cool, or sucks that much power ?

I guess fanboys never die.......

It's fanboyish to take processors and oveclock the hell out of it?

Btw dmens, they were mentioning watercooling a whole bunch but it sure as hell didn't sound like it. I could hear it halfway up the auditorium.. lol

 

[dmens]

Yeah man, no way that thing was watercooled. The thing woke me up when they turned it on, 2nd row from the back, haha.

I can imagine people buying that. Namely, people who are too lazy or don't know enough and want a watercooled setup factory guaranteed from Dell.