Intel Demos New P4 'Extreme Edition'

typobox43 writes "Louis Burns of Intel displayed a "high-definition video stream running on a 'mystery' desktop processor." This processor turned out to be the new Intel Pentium 4 Extreme Edition 3.20 GHz, with an extra 2 Megabytes of cache."

Comments

Possible Advertising Campaign?

[The_Rippa]

Saturday. Saturday! SATURDAY!

At Intel Headquarters!

Witness the unveiling of the next...

Biggest!

Meanest!

Fastest processor you can imagine.

Pen-Pent-Pentium EXXXXXTREME

It's 3.2 gigahertz of binary badness.

Come witness as it peforms calculations at mind-boggling speeds!

Special Guest The Blue Man Group

Tickets start at $20 for adults, discounts for children and seniors

If you miss this, you'd better be dead... or in jail...

And if you're in jail, break out!

Re:Possible Advertising Campaign?

[L-Train8]

Extreme? Whenever I hear that word in an advertising campaign, I think of Homer Simpson as Poochy the Pooch: "Hey kids, remember to recycle... to the extreme!"

Re:Possible Advertising Campaign?

[m00by]

dude, you so TOTALLY forgot the most important part: you pay for the whole seat, but you'll only need the EDGE!!! =D

Re:Possible Advertising Campaign?

[nrmrvrk]

Special guess celebrity: Randy Macho Man Savage:

Kid: "Pentium chips aren't an extreme sport, Macho Man..."

MM: "PENTIUM CHIPS NOT EXTREEEEEMMMMEEE!?!?! OOHHHH YEEAAAAH!!!!"

Re:Possible Advertising Campaign?

[MrLint]

Pentium Extreme! its just as good as having a 64 bit CPU!

Re:Possible Advertising Campaign?

[Anonymous Coward]

Pentium EXXXXXTREME*

*extras Xs may result from occasional floating-point errors.

Re:Possible Advertising Campaign?

[GMFTatsujin]

Change "pay for" to "license" and you've got a deal!

Re:Possible Advertising Campaign?

[Ibag]

When I hear the word extreme in an advertising campaign, I think of this. God bless Maddox.

Re:Possible Advertising Campaign?

[Doktor Memory]

...powered by SURGE!

Re:Possible Advertising Campaign?

[innosent]

...Pentium Extreme is much better than a 64 bit CPU, especially for the Pentium Extreme target market.

What, you mean for people who run 32-bit code?
The cache is something that's been sorely needed though, maybe Intel's finally going to make decent processors for people who need to run more than Word and Madden 2004. The problem is, they're stuck with the x86 instruction set architecture, which is crufty at best. Maybe once people start buying Itaniums (if ever), they'll get better.

Tips for Intel designers:
1) Get rid of 20-stage pipeline, it's too long for anything serious.
2) As a follow up to 1, try to actually get some work done in a clock cycle.
3) Throw out the x86 ISA.
4) Look at the MIPS ISA.
5) Realize that it's actually possible to understand the MIPS arch, and that it still works great for multimedia, math, and general use.
6) Buy the rights to the MIPS ISA, make small improvements (get rid of branch delay slot, load delay slot), speed it up, and design new Intel processors from the improved ISA.
7) Release versions of processors with 4MB Cache (2MB each I$, D$) for consumers, and 24MB Cache (8MB I$, 16MB D$) for servers/clustering/etc.
8) Release Motherboards for 1, 2, and 4 CPU configurations.
9) ...
10) Profit!

Re:Possible Advertising Campaign?

[grolschie]

err... dude. It doesn't matter how much cache they strap to these processors, deep down you'll still know it's a P4. :-(

Re:Possible Advertising Campaign?

[cheezedawg]

1) Get rid of 20-stage pipeline, it's too long for anything serious.

No its not. In fact, according to this research, the P4 pipeline is not deep enough. That paper concludes that P4 performance could be improved by up to 90% by increasing the pipeline depth to around 50 stages and increasing the cache size.

Do you actually think that Intel didn't know the consequences of increasing the pipeline depth? The Intel engineers didn't just guess on the P4 architecture- it was a very deliberate design decision. Judging by the P4's performance gains, it was a pretty good decision, too.

Re:Possible Advertising Campaign?

[mrm677]

1) Get rid of 20-stage pipeline, it's too long for anything serious.


No its not. It enables a high clock rate and with good branch prediction and selective replays, it is just fine.


2) As a follow up to 1, try to actually get some work done in a clock cycle.


Read some studies about the available ILP (instruction-level parallelism) in common applications. There really isn't much of it unless instruction windows are made huge which isn't feasible. This is why simulaneous-multithreading (hyperthreading) made it into an actual chip because it takes advantage of low ILP.


3) Throw out the x86 ISA.


Pentium4 is the leader of SpecINT. Not AMD, not Sun (RISC), not IBM (RISC), not MIPS (RISC). Some players, such as MIPS, don't have the resources to compete however IBM does.

And look whats happening to Itanium? Disaster even with its oh so elegant ISA.


4) Look at the MIPS ISA.


What about it? Yes, very clean and orthogonal. Intel and AMD have proven that an ISA is irrelevant to achieving high performance.


5) Realize that it's actually possible to understand the MIPS arch, and that it still works great for multimedia, math, and general use.


Realize that undergraduate computer architecture is simply an introduction.


6) Buy the rights to the MIPS ISA, make small improvements (get rid of branch delay slot, load delay slot), speed it up, and design new Intel processors from the improved ISA.


Unnecessary. Besides, I like software compatibility and binary translation just doesn't work at this level.


7) Release versions of processors with 4MB Cache (2MB each I$, D$) for consumers, and 24MB Cache (8MB I$, 16MB D$) for servers/clustering/etc.


2MB each for I$ and D$? Then you must be referring to L1 caches which are the only ones typically separated into data and instruction. And how many cycles would it take to access those caches?


8) Release Motherboards for 1, 2, and 4 CPU configurations.
9) ...
10) Profit!


Read Intel's annual report. They are quite profitable already and don't need the advice from someone with a B.S. in EE or CS and marvels at how great the MIPS ISA is.

maddox influence?

[edrugtrader]

they must be reading maddox's site

I have a 3.2Ghz PC that I bought for home...

[Osrin]

... I struggle to tax it with anything I do, including some of the more intensive games.

This "extreme" version of the chip has to be aimed at a very niche market, at least for the next couple of years until more processor intensive software catches up.

Re:I have a 3.2Ghz PC that I bought for home...

[Fulcrum of Evil]

This "extreme" version of the chip has to be aimed at a very niche market

Yes, it's the 'mine's bigger' market, though I wouldn't call it niche, exactly.

Re:I have a 3.2Ghz PC that I bought for home...

[Rosco P. Coltrane]

I struggle to tax it with anything I do, including some of the more intensive games.

Try "nice -n 20 dd bs=1 if=/dev/zero of=/dev/null" ...

Re:I have a 3.2Ghz PC that I bought for home...

[jawtheshark]

Uhm, I dunno about you.... But if my compiles take more than a minute, I just go out and take coffee... chat with the cute gal from marketing and eventually pull some practical jokes on my coworkers.

Of course, other people actually work at work.

Re:I have a 3.2Ghz PC that I bought for home...

[Anonymous Coward]

According to research at an English university, people will always buy whatever the guy in the shop says is "best".

Re:I have a 3.2Ghz PC that I bought for home...

[be-fan]

Try compiling KDE. My 2GHz P4 struggles to do it in under a working day. Heck, it takes nearly a minute just to recompile a KDE theme after making a change to it!

Re:I have a 3.2Ghz PC that I bought for home...

[Mitchell Mebane]

Try MPEG-4 encoding. With all the XviD settings at max, I get ~4 FPS on my Athlon XP 2400+ when transcoding DVDs at full 16:9 resolution.

Re:I have a 3.2Ghz PC that I bought for home...

[Glonoinha]

All a matter of time.

I remember reading once in a Usenet thread - some guy was trolling and asked 'will my P90, overclocked to 100MHz, be enough to handle the flight combat simulators you guys are discussing?'

The first time I read it it was hilarious because he was either bragging or dreaming, the P90 chip was out in limited supply at the time and was easily 50% faster than the common P60 machine used by the sim-gamers, not to mention the overclocking it. Of course it was going to be fast enough.

The second time I saw it (a few years later) it was hilarious because the bare minimum system for any sim/game was a PII/300 with a 3D graphics card and his P90 was so pitifully underpowered it didn't have a chance.

So we get to enjoy the 'is this CPU enough' question twice, generally, for any given CPU. Just a matter of timing.

Obligatory Waynes World

[Znonymous Coward]

Extreme close up! Whhoooooooooooo... Whhoooooooooooo.

Level Three Cache

[Master Bait]

Ho hum. I suppose if it was level two cache, Intel would have said so very loudly, so they just call it 'cache'.

Re:Level Three Cache

[Anonymous Coward]

Don't knock it, even if its operating at 1/3 the cpu bus, an extra 2MB of level 3 cache will give a significant boost to things like video games and many other interactive cpu intensive applications.

Re:Level Three Cache

[philthedrill]

The tradeoff with cache hierarchy is access time on hits vs. size.

You could increase the size of your L2 (or L1 for that matter), but you do this at the risk of sacrificing cycle time. This is part of the reason the P4 has such a tiny L1 D-cache.

With clock speeds climbing higher, the amount of time for a signal to traverse across a chip is no longer trivial, so retrieving data within N clock cycles is unrealistic with a large cache.

To add to that, the benefit over an L3 hit (even though it's much slower than L1 or L2) is that it's still much faster than main memory. DRAM is built for capacity.

Adding cache is somewhat of an easy way out in terms of adding performance with your transistor budget. You keep your power density reasonable and you also don't change the microarchitecture.

In conclusion, I think drugs are bad. The end.

Multiprocessor?

[tinrobot]

The rumors are that this chips are the same or very similar to the $4000 Xeon MPs with 2MB cache. I wonder if these will work on the workstation class MP motherboards. Would be sweeeeet.

Re:Multiprocessor?

[loopWork]

Different pinout, so no.

Ohpleaseohpleaseohpleaseohpleas

[JoeLinux]

I wanna see major competition between Intel and AMD. That way I can get my 875P motherboard "tossed in free with the purchase of any Intel Pentium 4 Extreme(tm) Processor." It's about time I upgraded from a Celeron 433 anyway. Ghost Recon plays more like Ghost Recon: The Slideshow.

Joe

Tom's Hardware reviewed a similar Xeon...

[tugrul]

... with 1MB of L3. The results weren't that exciting.

Re:Tom's Hardware reviewed a similar Xeon...

[ciroknight]

But at the same time, the Level Three Cache is MUCH "further away" from the core in the sense that it takes much longer for data to travel accross the lines of the processor to get to it. Level Two isn't much closer, but that little edge does make a huge difference in this case. Game developers now have room to seriously push their applications because the processor will be able to cache more (data||instructions). It should vastly improve scores on very memory intensive apps.

On the other hand, I would much rather see them quadruple the size of the Level One Cache. This would improve performance on these processors, but at the same time, without the extra registers that a 64-bit chip would have, these improvements are limited by their usefulness, not to mention they would take up loads more valuable core real estate. I can't wait to see Intel move to a 64-bit chip with a 2 meg level 2 and maybe a 128k level one... we'd start to see chips FLY....

Re:Tom's Hardware reviewed a similar Xeon...

[akuma(x86)]

Computer architecture 101.

Average memory access latency per memory access =
(L1_hit_rate * L1_hit_cycle_time) +
(L1_miss_L2_hit_rate * L2_hit_cycle_time) +
(L2_miss_L3_hit_rate * L3_hit_cycle_time) +
(L3_miss_rate * DRAM_latency)

80-95% of your accesses will hit the 8k L1 in typical applications. This is the vast majority of the accesses. The latency of this cache is TINY on a P4. Do the math for a 3.2GHz 3 cycle cache.

Given a curve of cache-size vs. latency and hit rates for all the cache sizes, the optimal hierarchy is a simple optimization problem. I can assure you that this equation has been solved and the optimal heirarchy has been chosen (given the other constraints of obviously die-size and power).

Quadrupling the L1 will double the latency and kill your average access time, making your chip almost certainly slower.

Bigger caches mean longer latencies. It's limited by the basic laws of physics. There's only so much distance you can traverse in a ceratin amount of time and larger caches have longer distances (meaning higher RC delays).

The reason we want larger outer level caches is because the DRAM_latency is enourmous and has an impact on average access time. Hardware prefetching can also help to alleviate this problem - This solution is available on both Athlon and P4 chips and will only get better in the future because it is absolutely critical to hide this DRAM latency.

Ok, now to address the notion that more registers will improve performance...
You won't get as much performance out of more registers as you might think. First of all, when the compiler runs out of registers it spills the excess to the stack -- pushing it out with a store (spill) and reading it back in with a load (fill).

In modern processors (just about every chip out on the market), there is the concept of store buffers. Each store writes it's data to a store buffer. Subsequent loads that require data from stores, get their data by forwarding out of the store buffer. So -- the spilled store writes the buffer and the fill load reads the buffer -- all of this happening much faster than a memory access because it's just reading out a local on-chip buffer, so the load looks more like a fast register read. This architectural trick emulates the effect of having more registers, subject to the size of your store buffer. There are even more advanced architectural tricks you can play to completely eliminate the spill-fill pair from the critical path (look up memory-renaming in the literature).

If you're worried about chip-real estate, you should be very concerned that a 64-bit application's pointers will take up twice as much space effectively making your caches and memory bandwidth appear smaller.

Re:Tom's Hardware reviewed a similar Xeon...

[akuma(x86)]

Lets look at some numbers...

What is the standard time slice quantum for windows and linux typically? That is to say, what is the typical rate of context switches? If I recall correctly, it's on the order of 100 per second.

That's 1 context switch every 0.01 seconds. Lets suppose now that I have a typical P4 system with 6.4GB/sec of memory bandwidth. I can fill the entire 2M cache in roughly

0.002/6.4 seconds = 0.0003125 seconds

That's only 3% of the entire time slice quantum! That's assuming the thread will want all of the cache (which is unlikely for many apps).

So, yes you may be running 25 programs, but they only switch between each other 100 times a second. The cache re-fill only takes 3% of a minimum quantum, leaving 97% of the time left to read the filled data out of your caches (which is highly likely due to the principle of locality and LRU replacement policies).

Now, if you're running something server-like TPCC with a bazillion threads, then you'll probably context switch a lot more, but the P4 isn't designed for transactional servers. This is why servers systems have larger caches and multiple processors and multiple SMT threads within each processor -- among many other reasons :)

Putting the DRAM on die is a nice idea, but it's not cost effective at this time. You only have a finite die size. The cost of the die increases exponentially with area, so there's a hard cap on how big you can make the chip. The area that you would have spent on the DRAM could be used more profitably on other chip features. Perhaps in the future when we get many more transistors, it may be feasible.

text incase of /.ing

[Anonymous Coward]

Intel Developer Forum Cache for questions

By Nebojsa Novakovic: Tuesday 16 September 2003, 18:14
WHEN, AT today's IDF opening, Louis Burns demonstrated a high-definition video stream running on a "mystery" desktop processor, everyone must hve thought it was the upcoming Prescott part. Wrong! It was the (also upcoming), previously unheard of, even at The Inq, Intel(R) Pentium(R) 4 processor Extreme Edition 3.20 GHz , with an extra 2 Megabytes of pron. In Intel's own words, "this new processor will be targeted at high-end gamers and computing power users."

As a matter of fact, 2MB cache will help a lot those users whose apps (including games and such) have a lot of big cache-friendly *wink* pieces of code and data, but probably not the data-streaming intensive stuff. I do expect to see speedups anywhere from 2% to 20% depending on the application, maybe some more if using multithreading/multitasking (large cache can keep in code / date pieces from more threads).

However, this doesn't seem to be a new CPU in reality - after all, Intel is doing very well with its XeonMP 2.8 GHz 2 MB cache CPU, and how much effort does it really take to repackage it for the 3.2 GHz / 800 FSB desktop with less stringent thermal and reliability requirements than the big iron, anyway?

Intel would gain a lot with this move. If, touch wood, there are problems with Prescott, a large-cache Pentium4 part will provide some buffer against large-cache Athlon64 (i.e. rebadged Opteron) parts. At the same time, enormous extra benefits from the economies of scale would further reduce the identical die XeonMP manufacturing cost, helping Intel compete better on the quad-CPU server front as well. Interesting move? I think so. Let's see how the beast performs in real!

More impressed with AMD.

[Thinkit3]

What I'm really not impressed with is Intel saying desktop users don't need sixty-four bit. Well, we don't need gobs of cache. We need sixty-four bits.

Re:More impressed with AMD.

[C. Mattix]

The difference is that all the existing apps would need to be recompiled to fully use the 64bit. Even lowly DOS can use performance improvements with a larger cache. And with Hyperthreading the number of clocks per instruction is very small, this lends itself to using a larger cache more often.

See also:

Ars Technia on Caching

Extreme price...

[Anonymous Coward]

$740 in 1,000 unit quantities. I think I'll pass.

Cinema-like video

[Rosco P. Coltrane]

Louis Burns of Intel displayed a "high-definition video stream running on a 'mystery' desktop processor.

Gosh, one of these days I'll have to take a sneak peak at the hardware they run in that mystery little room in my local theater. The monitor is so big, the soundcard is great, and I can see it all for a buck!

I will not buy anything that has X, eXtreme,

[teamhasnoi]

X-treme, XXXtreme, X-tream, XT-ream, AXEtreme, Xtreme, or is generally Xed-up in anyway.

Please send a message to the X-tra stupid Advertising XX-cutives that X in the name is X-tremely dated and not an X-ellent idea.

The new marketing buzzword is 'Shit-Hot', as in "The new Intel Shit-Hot P4!"

Thanks.

Paper Launch?

[Dumass]

Not to be labeled a fanboy (although not necessarily denying that status)... but this sounds like a paper launch just to take some press away from AMD.

"He [Burns] said the chip will be available to buy in the 30-60-day timeframe." from this article.

Prescott is going to be late and has been getting bad press for not being backward compatible with current motherboards. Why not make some noise with a product that wont be around for another month?

extra 2 mb's?

[mOoZik]

So it's not just a 2 meg cache but is in ADDITION to an existing amount? 256? 512? I'm confused.

Re:extra 2 mb's?

[Bored Huge Krill]

yes, it is in addition. There's a report at Anandtech that writes up some of the details

What is new is a 2MB L3 cache. It's made clear that this is in addition to the existing 512kB L2 cache on current P4s, making a total cache size of 2.5MB

database searches

[chipace]

This chip would be great for database searches... it has more cache than uni-processor xeons and it probably will be cheaper. Thanks gamers! I guess the wait for Prescott is real... seeing that Intel had this chip on tap.

CNET article with more details

[HeroicAutobot]

CNET has an article with more details (or speculation more likely).

Some interesting quotes:

"The performance boost is awesome," Burns said Tuesday during a speech at the Intel Developer Forum here.

"It is a Xeon with a different pin-out, or least that's what it looks like to me," said Nathan Brookwood, an analyst at Insight 64.

Intel did not disclose the price of the Pentium 4 Extreme Edition. It likely will be as expensive as its counterpart, the 2.8GHz Xeon with 2MB cache. That chip sells for $3,692 in quantities of 1,000.

"It absolutely will be kind of pricey," Brookwood said.

Wow!

[ENOENT]

3.2 GHz! That's 6.7% faster than 3.0 GHz! You feel the need to send money to Intel! Fnord! Imagine how fast the Internet will be if you have one of these on your desktop! You will need a neon-colored bunny suit just to look at your computer! You will be assimilated by the Blue Man Group!

Re:Wow!

[Aadain2001]

I remember the days when an increase in only 50 MHz as a big deal! Now they role at a 200 MHz increase and people say "it's only a 6.7% increase, big deal"?!?. What, do you expect them to role out a 1 GHz increase with each new chip they put out? Time for a little visit back to reality.

Re:Wow!

[dustinmarc]

Sure, Intel is releasing smaller percentage increments, they make more money from it. The nice thing is that they still follow Moore's law. So ultimately chip speeds still double about every two years.

Re:64bit vs 32bit

[2nd Post!]

Nope, the only 'desktop' 64 bit processors come from IBM and AMD;
AMD Opteron
AMD Athlon64
IBM PPC970

Intel's 64 bit solutions is the Itanium! Anything with the Pentium moniker is 32 bit. The Itanium is the one which suffers 32 bit emulation lag.

So if you want 64 bit, you're stuck with, realistically, a Mac or some brand of Athlon CPU.

I wonder if you can use it with BTX motherboard

[Ratface]

Would that give you a PC with BuTtoX Extreme inside?? :-D

Re:64bit vs 32bit

[Naito]

You're thinking of the Itanium. It used to run 32bit X86 under a hardware emulator, but that was about as fast as the Pentium MMX. Intel has since switched to using a software emulator, something like Transmeta does with the Cruesoe, and it's actually faster than the hardware emulator, about the same speed as a Pentium III now.

The Xeon is a Pentium4 in different packaging and with SMP enabled. Actually, SMP is probably enabled with the Pentium4 too, but since there are no such motherboards and you can't plug them into Xeon DP mobos, nobody can test that. Xeons already come in versions that have up to 8MB of L3 cache, the new Pentium 4 is probably just a rebadged Xeon certified to run on an 800Mhz bus.

Re:64bit vs 32bit

[chill]

You're confused.

The Xeon series has always been Intel's "server" chips. Mostly a different pin out and lots more cache. They're souped up versions of the normal chips.

The Itanium is the 64-bit unit.

breaking news

[colinleroy]

as usual from Intel... Just add some megahertz and some cache, it'll be good enough. On paper. What about improving technologies, like IBM or AMD do ?

Re:Interesting, but...

[militantbob]

I need that much juice. My Windows machine typically handles mIRC, Yahoo Messenger, 3D Studio Max, 3 or 4 IE instances, etc.

Of course, most of those programs are essentially idle at any given moment. But when I'm trying to render a massive 3dmax scene while switching over IRC to ramble libertarianesquely about the failings and dangers of big government, while at the same time opening/reading 3 or 4 web documents... my machine bogs down on me. Now, this machine is a P4-2.9GHz with a gig of RAM on SCSI disks... perhaps the extended speed and cache on the new CPU would make a difference.

At the same time, I could use one of these on my colo box, which is hosting 17 domains with about 3,000 pageloads per hour. Then again, I could always get something other than x86, if I weren't a cheap and ignorant bastard.

Processor-Intensive SW: Engineering Applications

[reporter]

This "extreme" version of the chip has to be aimed at a very niche market, at least for the next couple of years until more processor intensive software catches up.

The processor-intensive software is already here. It is called HSpice, Verilog, fluid-dynamics simulation, etc. The Pentium 4 has done nicely in the engineering workstation market, and the "Extreme Edition" should do even better.

Please check the SPEC web site for a performance evaluation of the Pentium 4's floating-point (FP) performance. In particular, it outperforms the UltraSPARC III even though the latter has a 2-to-1 advantage in the width of its databus -- 64 bits versus 32 bits.

What changed the x86 chips from also-ran losers in FP performance to the kings of the hill? SSE.

The SSE extension to the x86 instruction set architecture (ISA) opened up a whole new world of applications for the Pentium III and successors. Older Pentiums were saddled with a FP stack that hurt their performance. The SSE extension established a directly addressable bank of 8 128-bit registers or 32 32-bit registers for FP operations. As a result, the Pentium 4 outperforms the UltraSPARC III on video applications.

At 3.2 GHz, the "Extreme Edition" of the Pentium 4 should help the Pentium 4 to capture even more of the engineering workstation market. Nowadays, the first-choice workstation among engineers in Silicon Valley and Boston's Route 128 is Linux running on a fast Pentium/Athlon, not Solaris lumbering on a slow UltraSPARC III.

... from the desk of the reporter

Re:64bit vs 32bit

[Slack3r78]

The Xenon is 32 bit. You're thinking of the Itanium.

Re:I'm so sick of "extreme" this and "Xtreme" that

[kfg]

Well something had to replace "turbo."

I was so disappointed when I cracked the case of my 286 and didn't find an HKK in there somewhere. I had thought that maybe I could replace the one big one with two little ones to reduce net lag.

From the marketing point of view the advantage of "Xtreme" is that you can't prove it isn't in there somewhere. Maybe keeping the Magic Smoke in or something.

The disadvantage is that they can't play games by making you think the go slower button is really a go faster button. I hated having to explain that to people, they always got mad at me.

Killing the messenger always works, if you're a nitwit.

KFG

1GHz is plenty!

[Daniel Wood]

Yeah, you probably were one of those people too.

This CPU is aimed at the gaming/multimedia community. All that extra cache should make Doom3/HL2 speed along a little better. It should also help us that encode DVDs/DivX on the fly. What supprises me is that they didn't finally go to 1GHz FSB. Yeah, I know, that would mean you need DDR500(PC4000). While I'm sure you make have problems taxing your 3.2GHz CPU with MS Word or Counter-Strike, I am left longing for more CPU power with my Dual Athlon MP2100+ when De-interlacing video from my DV cam or running Urban Terror with SMP @ 1152x864x32+2xAA+AF.

You say software needs to catch up? I say hardware has and will be playing catchup for a long time. I'm sure that hardware will never exceed the demands of software. Multimedia has been the driving force behind computers since they became "good enough for Twin/QuatroPro/WP/etc." and will continue to be that driving force.

Re:1GHz is plenty!

[Pharmboy]

What supprises me is that they didn't finally go to 1GHz FSB. , I know, that would mean you need DDR500(PC4000).

Actually, no you don't. Apple sells their dual 2ghz box, that has a 1ghz fsb (dual pipe), and 400mhz ram. goto apple.com/powermac for info. It obviously doesn't talk to the ram that fast, but 1g pipe to the chipset doesn't suck either. Oh yea, and up to 8gb of ram so far. its a bit different in other aspects as well.

I am just waiting to score one of the dual 2.0 boxes used (cant afford $3500) but that will take a while. They also bench out better cycle to cycle that intel (similar to amd or better) Its actually the IBM 970 cpus (reduced power 4 cpu) that IBM is said to be releasing soon in entry level servers, with 4 cpus, for $3500, for Linux.

Re:I'm so sick of "extreme" this and "Xtreme" that

[mr.henry]

IANAAM (I am not an advertising major..), but apparently the "X" is supposed to make us think of sex, and therefore make whatever product a company is pitching more appealing. "SX" is even more blatant. In product model lineups, it's everywhere.

With that in mind, and seeing past the fnords, LX or LS (think Lexus LS 400, or whatever the latest is), is the most appealing of all: lesbian sex.

I hope I don't come across as crazy or perverted, but advertising will do ANYTHING to sell crap to people.

Re:Interesting, but...

[Graspee_Leemoor]

" I need that much juice. My Windows machine typically handles mIRC, Yahoo Messenger, 3D Studio Max, 3 or 4 IE instances, etc."

3 or (gasp!) 4!!!! instances of IE?!?!?!?

Dude, you are XTREEEEEEEEEEME!

graspee

Wait for the next version after this

[PanchoVilla]

In a couple of months Intel will be releasing the new Pentium IV TypeR !!!! The heatsink will even have one of those ugly shopping cart handle type spoilers and NEON too...........

EXTREME? How about stable?

[narftrek]

What is it with Extreme as the buzzword these days? When you hear extreme you think of people jumping off cliffs or launching motorcycles off tall things. Things that some may consider DANGEROUS or STUPID. It can also mean "on the edge" as in pushing the limits or ground breaking technology. I don't know about the rest of you but I don't want a computer that pushes the edge, is dangerous, or stupid. I want a nice stable (as in doesn't crash 10 times a day) computer that I can watch my pr0n on. Is that too much to ask? Extreme is worn out in my book-pick a new buzzword.

Re:Interesting, but...

[commodoresloat]

...of all the people who are going to run out and buy it, who REALLY needs that much juice?

Some of these people even believe they need more than 640K of RAM.

Part 2 of Article up now

[dubiousdave]

The second part of their article is here.

Who said the 2MB cache was L2?

[tugrul]

According to this ExtremeTech article about this cpu, its L3.

The gaming-optimized Pentium 4 contains 2 Mbytes of level-3 cache, and will work with existing "Springdale" and "Canterwood" chipsets, Burns said.

Re:*sigh*

[smart.id]

I'll tell you how Intel stays in business. I know of people who buy a new, state of the art computer every year. People think that you need to in order to "keep up" with the latest technology. I can understand why you're using a 800mhz computer if it works fine for you, I have a 1ghz T-bird that has lasted me 3 years, and I don't need anything better. However there are people who feel they do, and also need to buy every single type of PC possible (tablet PC, pocket PC, laptop PC, desktop PC etc.)

Awwww

[sys$manager]

I was hoping for Pentuim 4 Turbo Alpha.

HAAAAADOUKEN

Re:Interesting, but...

[wondafucka]

I can think of an example:

If you are using a PC to emulate a real time system, that cache would help out in keeping your processes from having to access the slower memory. That means a few extra operations per unit time.

More specifically, if you are doing real time wave modeling that would mean a slightly more complex waveform. Shure you could use a DSP, but if you are in the development stage, it's usually easier to use a desktop system.

Another example would be if you are approximating physical hardware at high speeds. That's a few extra logic gates.

I'm not going to buy one, but if I had the money I would consider it.

Oh no, not again

[jazman]

Here we go again. Fastest chip in the world; nobody, not even weather computer people could ever want faster, blah blah, bollocks bollocks; everyone knows in 12 months granny will want a PC with one in it and we'll all "need" something with a gazyllion terabytes of RAM and that runs at a googolplex hertz just to do some silly emails and stuff.

Was looking through a 1984 copy of Personal Computer World and it was saying exactly the same about the new 2MHz 8086 or whatever. Would've thought those crazy marketers'd have learnt by now that in IT there's a new "fastest in the world" every few months.

Still, I suppose some people will be new around here and be impressed by this sort of crap. I know I was, first time (probably in 1981 or thereabouts) I saw a front page "Fastest in the World!" story; second time round I thought hang on, haven't we been here before?

And yes, I know that's before some of you here were born, before any of you point it out and depress me even further.

Am I missing something?

[WndrBr3d]

Perhaps I'm confused here, but I remember TomsHardware doing an article on the new Barton processors with double the cache (512k) didnt produce really noticable performance increases in most 'high end user' applications (gaming/video encoding.

Could Intel be planning a compiler that would utilize this cache??

Re:Am I missing something?

[Indy1]

the P4's archtecture (sp) is such that it is incredibly sensitive to cache misses due to its long
pipeline (20 i believe). Thats why higher memory bandwidth and larger caches make such a huge difference on the P4. Where as the Athlons have a much shorter pipeline (12 i believe), the extra memory and cache dont help out as much.

Re:Processor-Intensive SW: Engineering Application

[Hoser McMoose]

In particular, it outperforms the UltraSPARC III even though the latter has a 2-to-1 advantage in the width of its databus -- 64 bits versus 32 bits.

Err.. The P4 has a 64-bit data bus. The UltraSparcIII has quite a different databus (due to it's integrated memory controller), but when you look at memory bandwidth, the USIII has 2.4GB/s of memory bandwidth while the P4 has 6.4GB/s.

What changed the x86 chips from also-ran losers in FP performance to the kings of the hill? SSE.

Less than 5% of SpecFP scores make use of SSE. The performance comes mainly from the P4 having a lot of memory bandwidth. The only chips with more memory bandwidth are the Alpha 21364, the Power4, the Itanium2 and the Opteron. Ohh, take a guess as to which chips get higher SpecFP scores than the Pentium4 does.

Where's the dealcloser?

[contrasutra]

Well, im EXTREMEly disappointed.

If i'm an uber-kewl l33t gam3r k1d, and i'm buying this just to show my equally pathetic friends that im the 0wn3r, how am I suppose to do this without a TEESHIRT!! I NEED TO SHOW OFF!

Intel may never get their market.

Re:I'm so sick of "extreme" this and "Xtreme" that

[questamor]

from thesaurus.com:

Entry:
extreme

Synonyms:
acute, consummate, great, greatest, high, highest, intense, maximal, maximum, severe, sovereign, supreme, top, ultimate, utmost, uttermost

Antonyms:
limited, mild, moderate

"P4 Acute" - sounds like a Honda
"P4 Consummate" - sounds like something you'd cook
"P4 Great" - sounds retarded
"P4 High" - sounds like a urinal competition
"P4 Intense" - sounds like a pun
"P4 Maximal" - sounds like a condom
"P4 Sovereign" - sounds like an archaeological discovery
"P4 Severe" - sounds like a poisons classification
"P4 Supreme" - sounds like a pizza
"P4 top" - sounds like a urinal competition to avoid
"P4 ultimate" - sounds like more condoms
"P4 utmost" - sounds dumb

I'll settle for P4 Extreme

Re:Processor-Intensive SW: Engineering Application

[mczak]

SSE is single-precision (32bit) floats only, so pretty useless for scientific calculations (usually require doubles).
However, I believe the intel compiler uses SSE2 (which can handle 64bit floats) exclusively for float code, since the P4 legacy fpu is just slow. Of course there are compiler switches for the compiler so the code also runs on good old Athlon, Athlon XP, PIII (which lack SSE2, the Athlon also lacks SSE) - and those aren't exactly slow doing float calculations neither.

Faster CPU lengthen useful life of machine

[AHumbleOpinion]

This "extreme" version of the chip has to be aimed at a very niche market, at least for the next couple of years until more processor intensive software catches up.

While I agree with one of the other posters that many high end CPUs are sold to the "mine's bigger" crowd, Intel naming surely supports this idea, there are some legitimate advantages to getting a faster CPU even when you don't have a need for the additional computational power. I'm getting along well with a P3 1.2G but towards the end of the year I will be building myself a new machine. A P4 2.26G 533FSB would be fine but I'll put together a 3.0G 800FSB dual-channel DDR because it will only have a relatively modest price increase (then, not now) but it will add a year or so to the useful life of the machine. For years I've had the same strategy. A high quality motherboard for US$150 or less, the fastest CPU for US$250 or less, the largest HD for US$150 or less, ... My systems are not the fastest out there but they are close to the price/performance sweet spot and have good longevity since they are far more than what I need at the time.

Re:64bit vs 32bit

[Brandybuck]

You're both wrong. Xenon is a noble gas. Xeon is a 32 bit processor.

Re:EXTREME? How about stable?

[shadowcabbit]

Being an excessive literalist myself, I always think that "extreme" indicates that something is at the far end of a spectrum.

For example, this new chip is at the far positive end of the price spectrum, and at the far negative end of the "will I really need this in the next three years" spectrum. It, being on the far ends of two spectrums, qualifies as EXXTREME.

(Nevermind that my first online nick had xtreme in it. I was 15, sue me.)

Old IBM XT?

[WatertonMan]

Wasn't the 80286 in the old IBM XTs the "extreme" chip. At least I thought that was what the XT stood for. Maybe it stood for extra? Anyone know?

Re:Old IBM XT?

[shadowcabbit]

On my friend's computer it seemed to stand for "Xcruciating Torment". He hated that thing.

Re:Databus of Pentium 4 is 32 bits, not 64 bits.

[Arker]

He already explained it, it's not magic, it's mostly memory bandwidth. The size of that internal bus doesn't mean squat when it's sitting waiting for data from main memory.

Re:Databus of Pentium 4 is 32 bits, not 64 bits.

[Hoser McMoose]

Err, no. The internal data bus of the P4 is 256-bits wide, at least if you're talking about it's L2 cache bus. L1 cache doesn't really have a "bus", especially not the P4's trace cache (it's replacement for an L1 i-cache), but if my memory serves me correctly, the L1 d-cache of the P4 can read or write a pair of 64-byte values in 2 clock cycles. I guess that makes it's "bus" 128 bytes (not bits) wide. I don't know the bus width of this new L3 cache on this P4 "Extreme", aka a XeonMP, but I would guess it's 64-bits wide.

I haven't got a clue as to the internal data bus of the USIII, but I would guess that it's either 128-bit or 256-bit wide. Side note: the Power4 uses a MASSIVE 1024-bit wide internal bus, one of the reasons for it's impressive performance.

The only situation where the USIII has 64-bits and the P4 has 32-bits is if you are talking about integer registers or memory pointer width, neither of which are going to play a role in Spec CFP scores.

Apples compared with Oranges

[zealotasd]

You are making a incorrect comparison in computing technologies. UltraSPARC III is for higher precision, but it is way out of its competitive market by two years ago. Pentium 4 is built for highest performance at the expense of power consumption. In a more objective comparison with the UltraSPARC III, we would compare performance/initial cost/power consumption (and forecasted power consumption cost to price barrier). UltraSPARC III is built for good performance on its implemented hardware, thus it utilizes its bus and memory architecture to optimum. The Pentium 4 does not perform with the mathematical precision and architecture efficiency as does a UltraSPARC III. The Pentium 4's memory architecture isn't even being used to full efficiency because of the nature of x86 being a pro-legacy architecture.

The biggest black sheep of the industry is the legendary Alpha architecture. It's a 100% 64bit precision platform with highest efficiency per watt and it was purposely bought by Intel to be silenced and migrate all its users to the Itanium architecture. Not even an Itanium2 can perform as well as an Alpha of two years ago (21264/ev6). The only downfall of Alpha is the legitimate and objective comparison of performance/initial cost as being the notion it is highly non-competitive with other offers. The reason it is not as competitive with other architectures is not based on fabrication costs: it is based on it being the better architecure that was purchased before its parents' bankruptcy (DEC...Compaq?), and to try to recover the R&D costs of the overly-invested lesser architecture known as Itanium.

People who still use Alpha already know that if it is buried then the only logical successor would be a Power4 hands down. All the while, HP's PA-RISC is being incorporated into the same Itanium architecture to migrate its dwindling userbase to Itanium. So much is going wrong in the idustry it makes me sick to the stomach.

And here I was..

[bob670]

being all pleased with my 2gHz Celeron laptop I just got. I know there's a market for this, but it really strikes as just marketing to piss on AMDs 64bit parade (kind of like how nVidia always conveniently had a new Deotonator release that would boost performance 29% every time ATI released a new card, back when nVidia was on top performance).

Re:Speed / Cache is irrelevant *soon*

[darkwiz]

In the next year we'll see the first solid state hard drives (Some that will run fast or faster than the processor) and faster RAM that would run the same speed as the processor.

Cache on a processor would be redundant if you can access the RAM at the same speeds. AMD is aware of this and are working to make compatible products.

No.... we won't. What you are describing is insane. Come on: 3.2GHz x 32 bits? Access/transfer times over a full scale bus with a latency in picoseconds? Um... no.

There is a reason no one has done that yet - made system RAM the same speed as the CPU - and it ain't economics: it is physics. Nature does not take bribes.

Look, it isn't that it is too expensive to make fast RAM. And it isn't the distance - it is the capacitance. The problem with fast RAM is getting that signal off chip to the CPU. And the wires that connect the RAM and CPU are orders of magnitude higher capacitance than the wires on chip. That is a fundamental problem which you won't overcome without a fundamental change in how you move the data around.

Solid state drive/memory that runs at compatible speeds as the processor will probably reduce the need for what we call ram these days and operating systems could just use the drive for it's RAM.

Um.. no. Never will that be the case except in situations where using an archaicly small amount of processing power is adequate. Storage technology, as it is formulated now, cannot approach the speed of access, communication, and storage that even a low grade CPU would use for cache.

Maybe - MAYBE when we are using diamond wafers, high-temperature-superconductor-nanotube-quantum-d ot wires, and other buzzwords.

I saw this once before...

[lhpineapple]

It started off with Street Fighter 2. Then Street Fighter 2 Champion Edition. Then Street Fighter 2 Turbo. Then Street Fighter 2: The New Challengers...

I think it took them 3 or 4 years to actually get to Street Fighter 3.

*soon* is a decade or more away

[MarcQuadra]

I'm sorry, but I get a strong feeling that you're way out in dreamland.

Let's get some things straight first, no hard drive or ram can be 'as fast' as a processor, because that's like saying my coffee cup is as fast as my bicycle, it's meaningless.

A solid-state hard drive has to get all that data addressed, and it has to pump it over some sort of pipe at least several inches long, the addressing will put a buttload of latency in there, and that pipe would bring the bandwidth WAY down.

Now I wouldn't fuck with a solid-state drive at the end of a Ultra320-SCSI pipe, but that's STILL 1/10th the memory bandwidth of a modern DDR400 system. BTW, those have been around for AGES, I used to have a solid-state SCSI drive on my old 25MHz Mac, it was pretty fast, but nowhere near the internal ramdisk's speed.

RAM now works 'at the speed of the processor' if you think about it. My Athlon can chew about 2100MB/sec which is EXACTLY output of the memory I'm running (that's the 'sync' in SDRAM). The only way to change that would be to 'widen up' the CPU FSB. You could put single, dual, or quad-channel memory on your Athlon and it wouldn't make a dime's worth of difference in any benchmarks, the back of the chip is the limit, and current RAM meeets that need.

It's general knowledge that the more storage you can arrange for, the more complex your addressing system has to be to keep it tamed. Here's an example:

When your CPU asks for something it needs from RAM it asks for the contents of a block of memory, whose address is held in a pointer that is dynamic, but readily available. The CPU just 'gets' it. It's even better if that block is already in the cache, as the cache buffers will satisfy the request before the memory controller even bothers to retrieve the block from RAM.

When your app needs something from DISK it has to send a request through the OS (in RAM) to do a lookup in the filesystem and give an address, which is shuttled over to the disk driver to fetch from the drive and back to a generic filesystem driver to present to the app. Should the filesystem not have that data cached it has to perform a complex lookup of where the hell that file actually is on the disk, often traversing several directory files. It's very complex.

What you're saying will eventually happen, but not for at least a decade. Someday we WILL drop the two-tiered approach to personal computing, and ther'll be 'unified storage' for running apps and storing files (like the palm pilot, but better) and it will be good. Until then we've not yet miniaturized the electronics enough to move over to that paradigm. I think nanotech/biotech will play a HUGE role in making the memory, cpu, and IO processor components small enough to run cool and unplugged.

Re:Speed / Cache is irrelevant *soon*

[gerardrj]

On-die cache will never be "irrelevant". Even if the main RAM system runs at clock speed (or faster), there is still latency in getting the signals from the CPU, to RAM and getting a response.
The difference could be as much as 10x.
A solid state "disk" (SSD) would suffer even higher latencies with all the command overhead and the several bus systems that must be traversed/translated.

We've been promised SSDs for years. The last time I saw SSD units that were of a usable size, and reasonably priced compared to rotational media was in the mid 80s when 128KB was a lot of storage. Of course, the mid 80s is also the last time I saw a desktop computer that ran the CPU, RAM and system bus at the same clock speed.

Re:Speed / Cache is irrelevant *soon*

[akuma(x86)]

It's amazing how people spew out assertions without first stopping to think.

The fastest velocity that any human has observed is the speed of light. Some theoretical physicists believe that this is an absolute upper bound on speed :) Last time I checked, that's 3e8 meters/second...

Or in chip dimensions, that's 30 cm/nanosecond. At 10 GHz (0.1 nanosecond cycle time), light can only travel 3 cm in a clock cycle!!! That's approaching the typical dimensions of a chip! That's an absolute upper bound on speed -- this is light traveling through a vaccuum. Electrons traveling through semiconductor material won't go near as fast. You need "on the order" of a clock cycle just to traverse a typical chip. The reality is that as we scale technology forward, it may take several clock cycles just to traverse a chip.

I think you need to stop and reconsider your assertions. Latency is here to stay...it ain't going away unless someone finds a way around the current laws of physics.

What we need to do is realize this and architect around it by trying to hide latency in various ways (caching/buffering, prefetching, etc...)

Re:Processor-Intensive SW: Engineering Application

[Moraelin]

Well, honestly, nowadays "it beats a Sun box" doesn't even say much. It's just marginally more meaningful than "it beats my old ZX Spectrum."

The Suns still struggle barely above 1 GHz, have a slow cache, and so on. It also doesn't help that they're still saddled with SDRAM memory, too. (At least in the case of the cheaper workstations, on a 32 bit memory bus too.) If we're talking programs that draw something, it also doesn't help that they're saddled with outdated _and_ overpriced video cards. And so on.

Even without SSE, there's no way in heck for that UltraSparc III to keep up with a P4. E.g., Sun's Java doesn't even generate SSE code, and it still runs faster on Windows than on Solaris. Go figure.

For all the BS about the advantages of 64 bits, the reality is that in 64 bit mode an UltraSparc actually runs _slower_. So be thankful that most of the apps for it (and certainly all benchmarks) really are compiled in 32 bit mode.

Frankly, other than a few PHBs, and a couple of people who think they're some form of resitance against Wintel if they buy Suns, the rest of us don't even consider Sun to still be in the race any more.

So yeah, your words about running Linux on an Athlon or Pentium reflect exactly what I'd say to anyone considering a Sun box: Get the cheapest PC that Dell sells, or build your own Duron system, install Linux on it, and there you go. You now have a Unix workstation, and it runs circles around any of Sun's workstations. Or, much as I'm no Mac fan, get a Mac. It'll be 64 bit, and based on BSD too.