Prescott = furnace? Lame CPU situation...

[Dragonprince]

The Feb 8th HardOCP review of a 2.8E Prescott talks about motherboard components getting up to 235F trying to deliver the massive current required by the new Intel baby, thats not the CPU at 235F its power components on the Motherboard!!! :eek: are you kidding?!? How long will a board last at those temps? The board was so hot it melted the plastic blocks put under it for air flow. I can already smell the fried capacitor law suits coming....sounds like the 1.3Ghz P3 all over again, to imature to be released but we have to do something about AMD64 so screw the customer....

I am looking at a new Intel or AMD rig but honestly im not happy with the Prescott and Im not thrilled about an Athlon64 either at this point. I think I might just stay were I am and buy a top shelf video card. Count me as truely disappointed in ALL of the CPU's currently being offered by Intel and AMD :(

[tomchong]

i have to say i agree with you, but future A64/fx's look VERY promising.

[SewerBeing]

the future does look promising, the prescott is just a chip out for beta testing I mean what better way to test a new chip then to release and sell it. I totally agree with you though

[ghostrider]

Why are you not impressed by the A64 chips?

[hUMANbEATbOX]

a64's are good, and so are northwoods.

but i agree with you, in general, its bleak.

[tom10167]

HardOCP + Tom'shardware = unbiased
HardOCP has been known to favor AMD, and Tom's has been known to favor Intel, I don't go to either site.

[SewerBeing]

I dont like the whole single channel memory thing with the athlon 64's (not the fx line) plus they are supposed to be bad overclockers.

[CrashOv3r1De]

Well from experience I must say that the A64 is very impressive and FX-51 is far more impressive than the Prescott.

I have tried the A64 3400+ @ 2.7ghz, P4 3.2C @ 4.1, and now I am about to test my 2nd fx-51.

I know a 3.2c is no prescott but it is actually better imho. Right Prescott is not promising at all but it sure can help you solve your condensation problems with the vapochill/prommy. Heck who needs heaters anyway...Also good for warming up a cold room on a chilly day...I dont mean to be a fanboy. For the past 2 years I have only used intel cpu's but now amd has really begun to shine..

In a few months intel should get its act togethor and hopefully the new 3.4E will be cooler. As soon as intel finished working on their dathon core they will start workign on the heat issues of the prescott. After a few revisions we should see these babies run much cooler -- or intel can scratch its plans for having a 4.0E released.

[Dragonprince]

Originally posted by ghostrider
Why are you not impressed by the A64 chips?

The whole single channel memory controller issue bugs me, that and the low processor speed. I want to see a dual channel Athlon64 at 3Ghz so I can justify buying it, 2.2-2.4Ghz leaves me uninterested and I dont want a refridgerator running next to me to get to 3Ghz....all that SOI bull seems like marketing at this point...:mad:

[Freddie]

I think that the prescott is the P4 willimate all over again. A more-stage pipeline allow less to be done per mhz (just like the early willimate p4's to p3's). A failed core. I was sure that the whole idea of makeing the pipeline bigger was so that it was able to run a higher clock speeds. But 2.8 come on. 2.6's or maby even a 2.53 are better. Like has been said heat, but there is one reason to get one. You could cook breakfast on it and use it a central heating!

[silentfire]

your waiting for a 3gHz A64?
since when did gHz mean anything?

[Dragonprince]

If Ghz doesnt mean anything then why is your 1700 OC'd ? Seriously I didnt OC my Barton 2500 for bragging rights...it needed it.

Athlon64's are merely "ok" at 2-2.4Ghz ....I dont think they will be worth the upgrade until they see dual channel and 3Ghz.

Prescott is to hot....probably why they gave it a name that rhyms with hot. I will wait until Prescott rhyms with cool before spending my cash on it.

[brakezone]

A 256 bit GPU running at 500mhz is pretty powerful... and for graphics works better than a 3ghz 32-bit processor.

The amd 64 has more operations per clock cycle than the socket A's. The whole idea behind X86-64 was to practically double the operations per clock cycle by taking it from 32-64-bits. The performance increase even on a 32-bit operating system has made them popular. With a 64-bit OS and a 64-bit program maybe you will appreciate the effort and hard work they have put into keeping up with your demand.

[Captain Newbie]

Prescott generally looks like a flop; even those Intel fanboys at THG think so.

[Gnufsh]

Originally posted by brakezone
A 256 bit GPU running at 500mhz is pretty powerful... and for graphics works better than a 3ghz 32-bit processor.

The amd 64 has more operations per clock cycle than the socket A's. The whole idea behind X86-64 was to practically double the operations per clock cycle by taking it from 32-64-bits. The performance increase even on a 32-bit operating system has made them popular. With a 64-bit OS and a 64-bit program maybe you will appreciate the effort and hard work they have put into keeping up with your demand.
The only case where enlarging the integers to 64-bit will be helpful is in programs that actually use 64-bit ints. There aren't that many that need to at the consumer level. There will be some improvements from x86-64 for programs that don't need 64-bit ins, but those will come from the 8 additional GPRs and the 8 additional SSE2 registers. Other performance improvements will come from architectural improvements over the XP (on-die memory controller, bigger TLBs, etc). The performance in a 32-bit OS has nothing to do with x86-64 and everything to do withimprovements over the K7.

[brakezone]

nothing to do with x86-64??? So they made it 64-bits for nothing and magically it being faster has ABSOLUTELY Nothing to do with the fact that its 64-bits... please!

I know, my post over-exxagerates the performance improvement and it being caused by x86-64. Thats not exactly what I meant entirely. So really in a sense your right, but look how many things a computer does that are 32-bit presently, 32 bit Z-buffers, 32-bit colors, 32 bit is used all the time... why? because thats what the OS is. If it was 64-bit programmers can double all those.

There are limits to what could be done with 16 and 32-bit that will be extended with 64-bit.

LOOK! i don't know what i'm talkin about:p

I just make this stuff all up LOL

[Agent_Mull]

Well, i would get the amd64. in theory they could use up to a few exobytes of ram. thats a billion billion bytes.

[PYROMANIAC]

Originally posted by Dragonprince
If Ghz doesnt mean anything then why is your 1700 OC'd ? Seriously I didnt OC my Barton 2500 for bragging rights...it needed it.

Athlon64's are merely "ok" at 2-2.4Ghz ....I dont think they will be worth the upgrade until they see dual channel and 3Ghz.

Prescott is to hot....probably why they gave it a name that rhyms with hot. I will wait until Prescott rhyms with cool before spending my cash on it.

it's probly oced b/c he wants to o/c..

you do realize that amd's do more calcs per cycle(9) compared to intel(6) right? dual channel memory -cough- athlon fx? -cough-

if you want amd to make a 3ghz chip then tell them to make a processor that runs as inefficiently as the p4

[Quailane]

If Ghz doesnt mean anything then why is your 1700 OC'd ? Seriously I didnt OC my Barton 2500 for bragging rights...it needed it.

Ghz does mean something on the same platform. Between two platforms it is a useless, stupid, and ridiculous comparison.


Athlon64's are merely "ok" at 2-2.4Ghz ....I dont think they will be worth the upgrade until they see dual channel and 3Ghz.

You obviously haven't looked at any benchmarks recently. Besides, the dual channel nets you at most 3%. That 3 Ghz AMD is like trying to wait for intel to come out with a prescot at 4.2Ghz stock.

[brakezone]

Boy do i feel like a d$mb@ss now!

I guess thats why they have 256-bit GPUS so that they can sit around and go... more bits doesn't mean #@!$#

Oh yeah well I have a 64-bit brain and I can understand it

[brakezone]

x86-64 is the arch that amd64 uses. Its backwords compatable with x86. Saying that being an amd64 running in 32-bit mode does not utilize 64-bits in any way shape or form is ridiculous. That would be like saying half the chip only works- when the arch x86-64 makes all those transistors and registers = 1 entity that is ready to execute instructions. Its like saying that 32-computers can crunch on seti just as well as a 64 of the same computers. Sure 64-bit doesn't help in every single thing, but it helps where it really counts- where 32-bit just can't possibly keep its pace.

There are lots of things in my mind that I could use 64-bits with as opposed to 32-bits. I could write a program that would shuffle cards faster! There are 23000000000000000000000000000000000000000000000000
000000000000000 approximate combinations in a deck of cards. As you can see you take a number like 64 and sure thats not a big number, but 64 can quickly become a very HUGE number of possibilities, just as you can take 52-54 cards and end up with a HUGE number of possibilities. more bits means EVERYTHING. The lenths by which programmers can design algorithms and arch is endless and someday we'll be 2.3E70 bits and NOTHING will have changed.

what if i wanted to know all of the different possibilities of different decks of cards... like what if i had 2.3e70 decks of cards and each one has 2.3e70 combinations... thats like a quadrillion quadrillion sextillion billion agoogle zillion with a trillion zeros behind it.:rolleyes:

[Krowa 02]

Wait till the new stepping on the prescott comes out. My hope is to have a 4ghz one on my ic7 by years end. If not then to bad ill keep the cpu or get an M0 and run this thing for a while.

[Silent Buddha]

Yep, D-0 should be out before summer.

Dragonprince have you even looked at reviews lately? I think you are still trying use megahertz to compare different architectures.

[brakezone]

nintendo was 64-bit along time ago... Where have we been=/

We're so far behind!

[tom10167]

Originally posted by brakezone
nothing to do with x86-64??? So they made it 64-bits for nothing and magically it being faster has ABSOLUTELY Nothing to do with the fact that its 64-bits... please!


Gnufish is one of the smartest members we have on the board, and you're new hear, you probably shouldn't mouth off to someone like that, do you just bust in to another country and make fun of their leader?

[Silent Buddha]

brakezone, just because A64s are also faster in 32-bit apps does not mean that it is because of the 64-bit compatibility. That potential has yet to be fully utilized. As Gnufsh said, performance gain only occurs 64-bit integers are used as the registers are exponentially increased. This is why John Carmack isn't too enthusiastic about 64-bit since he doesn't think he will make 64-bit binaries since they're not needed. All of it can't even happen if Microsoft doesn't release XP-64. The main performance increased of the A64s in 32-bit is due to the on-die memory controller from the elimination of the northbridge and the dedicated HyperTransport bus for I/O.

[OC Detective]

Yup a 64 bit cpu will not run a 32bit application any quicker than a cpu that is identical in every way except for being 32 bit. As silent buddha rightly states its the memory controller etc that are making the difference right now for the A64/FX right now. So these cpus are damn good now but when additional applications, drivers etc come out for 64 bit then there will an additional improvement.

[brakezone]

okay!

I beleive you guys:p

I just make alot of assumptions sometimes ya know.

I mean I really thought the data came in as 32 bit, and the processor itself would still use the double l1's to do calculation... That just shows I don't know a whole lot about processors. I mean I am just guessing stuff based on things that I thought I knew. Cause i'm trying to figure out how they work I guess.

Can anybody explain it with clarity. I mean does calculation take place on the l1's? Or is that just a place where data comes in and then it is worked on by transistors?

[brakezone]

I wasn't trying to make fun of gnufish... I said in the post right after his post that I believed that he was at the very least, basically right, also I said I didn't know what i was talking about and that I was just making stuff up.

Thats really not how i meant for it to come out as to be mouthing off I was just saying what I thought based on what i'd read. I assumed amd64 was faster because it was able to do more at once even in 32-bit mode.

gnufsh is the man he knows all about amd64!!!

[Gnufsh]

Originally posted by brakezone
okay!

I beleive you guys:p

I just make alot of assumptions sometimes ya know.

I mean I really thought the data came in as 32 bit, and the processor itself would still use the double l1's to do calculation... That just shows I don't know a whole lot about processors. I mean I am just guessing stuff based on things that I thought I knew. Cause i'm trying to figure out how they work I guess.

Can anybody explain it with clarity. I mean does calculation take place on the l1's? Or is that just a place where data comes in and then it is worked on by transistors?
Calculation does not take place in the L1, it's just another level (a very fast one) of cache. Numbers that are actually being added, multiplied, etc and their results are stored in very tiny, very fast memory. If you'd like to learn more, you could head over to howstuffworks.com, and look at their microprocessor article. If that is isufficient (and it should be), arstechnica has a very nice series about understanding the microprocessor:
Understanding Caching and Performance:
http://arstechnica.com/paedia/c/caching/caching-1.html
Introduction to multithreading:
http://arstechnica.com/paedia/h/hyperthreading/hyperthreading-1.html
Bandwidth and Latency:
http://www.arstechnica.com/paedia/b/bandwidth-latency/bandwidth-latency-1.html
Understanding the Microprocessor:
http://arstechnica.com/paedia/c/cpu/part-1/cpu1-1.html
Understanding pipelining and superscalar execution:
http://arstechnica.com/paedia/c/cpu/part-2/cpu2-1.html

Good Links about x86-64/64-bit on the desktop:
http://arstechnica.com/cpu/03q1/x86-64/x86-64-1.html
http://www.chip-architect.com/news/2003_09_21_Detailed_Architecture_of_AMDs_64bit_Cor e.html (to be honest, this one is over my head)
http://anandtech.com/cpu/showdoc.html?i=1815 (Anand's intro to the K8/Opteron arch)
http://www.xbitlabs.com/articles/cpu/display/64bit.html (Xbit Labs' take on 64-bit on the desktop)

Am I behind the times still callingthe instruction set x86-64?

And brakezone, don't worry you didn't insult me or anything. If I'm wrong I like people to point it out, so I can get it right next time. In this case, however, the integers being 64-bit alone doesn't give a performance advantage to most apps, even recompiling them to use 64-bit long mode. The programs that do use 64-bit integers will see a massive speed-up. For ones that don't a recompile will increase performance because of the extra registers available in 64-bit long mode (an SSE2). All programs will recieve a performance boost from the various improvements over te K7 core.

[XWRed1]

There is this also:
http://arstechnica.com/etc/linux/2003/linux.ars-01142004-2.html

(Can't help pimping it, I wrote it :P )