Thoroughbred Released Today [Archive] - Overclockers NZ Forums

Deviant

10-06-2002, 04:56 PM

Overclocking sux though, how disapointing.

Hopefully some new steppings will come soon that will make it better.

Reviews Here:
http://www.hothardware.com/hh_files/CCAM/amd_axp_2200.shtml
http://active-hardware.com/english/reviews/processor/xp2200.htm
http://www.amdmb.com/article-display.php?ArticleID=187
http://tech-report.com/reviews/2002q2/athlonxp-2200/index.x?pg=1
http://www.amdzone.com/articleview.cfm?articleid=1124
http://www.tomshardware.com/cpu/02q2/020610/index.html

_N_

10-06-2002, 05:03 PM

LOL, this is by far the worst OCer I've seen, unless they have the real batch of CPU's hidden somewhere, it's not gonna be very popular with OCers.

Deviant

10-06-2002, 05:08 PM

Yes, the Palimino is the better overclocker. I hope AMD sort this **** out soon, no wonder the T-Bred was released late.

And you can't use an aluminium base heatsink with it. The die is so small with a large heat output copper based HSF are the minimum to stop overheating.

Antallica

10-06-2002, 05:14 PM

I'll have to agree with ya _N_, those benches are budget as. I'll check against my benches for my 1600+ @ 1900+ ............. yeah I'm basically just as good. AMD's stuffed up here, hopefully with newer versions, steppings, etc of the 2200+ will be more acceptable of oc'ing.

Mabye they were gettin scared of Intel with their ever growing clock speeds and rushed a bit (just a thought) Not that high clock speeds would ever make me switch to Intel.

A big thumbs down to AMD for this toss up :mad:

Tiggerz

10-06-2002, 06:40 PM

Nadgers.. had it in my diary for tomorrw.

Just seems to look like a bug fix upgrade to me.. would have thought they would have attempted sse2 extensions. thats where the cpu falls down.. seems a shame to delay that til hammer stuff comes out.

Geek4Life

10-06-2002, 07:01 PM

What will be the point if they don't OC as well.

KingJackal

10-06-2002, 07:03 PM

Interesting - vveeeerrrry interesting.

I don't think AMD will have 'rushed' these chips out. Intel and AMD are too big and too smart for some dumb engineer to get the jitters and ship a product too early.

As for overclocking - that is VERY strange. Not because it's so low, but rather because of the chips these reviewers have! They all have 2200+ chips which don't overclock much, while stores are taking pre-orders on 2500+ chips. Waaaiiiittt a minute.....

LOL @ TOMS!! :p :p
He only got his to 1890MHz, while the AMDMB folks got theirs to 2160MHz stable - heh, pick the REAL overclockers.
LOL @ TOMS :p :p

Oh yeah, and almost 20% on a debut processor isn't actually low at all - some would call that pretty good ( bear in mind these reviews will have been prepared previously under NDA's - those guys' chips will be quite old compared to what the fabs are pumping out now ;) ).

KingJackal

10-06-2002, 07:04 PM

Originally posted by Tiggerz
would have thought they would have attempted sse2 extensions. thats where the cpu falls down..

Are they ALLOWED?
( bearing in mind SSE and SSE2 are proprietary APIs owned and designed by Intel.... :confused: )

Deviant

10-06-2002, 07:13 PM

AMD never said SSE2 would be in T-Breds, but have said all along that SSE2 would be in Hammer.

Good points you raised KJ, no doubt we will see some real overclocking over the next week or so.

Also the XP2500+ must be a totally different stepping, and so the current XP2200+ might be old stuff from a month ago that they just stockpiled. Maybe 2200+ on an 2500+ stepping would be a good overclocker.

SecretSquirrel

10-06-2002, 08:06 PM

Originally posted by KingJackal
Oh yeah, and almost 20% on a debut processor isn't actually low at all - some would call that pretty good ( bear in mind these reviews will have been prepared previously under NDA's - those guys' chips will be quite old compared to what the fabs are pumping out now ;) ).

hehe, I was reading all the negative comments and I hoped for someone to point out the fact that these chips will have been the batch thrown out to the reviewers a while ago...I suppose these chips will be the first to prove we're all going to have to go watery aren't they?

I agree with the 2200+ being 'older' stock
They probably gave them away to all the people who review, all the families of the company because they just dont do what the 2500+ will do ;)

Or something...

In conclusion : persons who are jumping the gun to throw the toys down at the new chip...HAVE FAITH, AND SOME PATIENCE, KEMO SABES.

mird-OC

10-06-2002, 09:09 PM

hehehe. i have faith, i'm buying one anyway :)

Volodkovich

10-06-2002, 09:31 PM

i think all we have to do is wait...the Xp1800+'s dont seem to be going too well, and i would have a guess that the XP2500+'s are going to be alot betta. Overclockers.com.au got there XP1800+ to 2.05ghz stable with a vapochill. A palomino would do that. Its probably just the first steppings, i mean, how well does the first stepping of a palomino OC?

theYANK

10-06-2002, 10:44 PM

Two things I'm worried about: that small a$$ core with NO heatspreader = OC'ing nightmare, and 266 FSB STILL with no plans to increase it even through Barton...come on AMD! The EV6 bus can be quad-pumped easily! :confused:

KingJackal

10-06-2002, 11:09 PM

Originally posted by theYANK
Two things I'm worried about: that small a$$ core with NO heatspreader = OC'ing nightmare, and 266 FSB STILL with no plans to increase it even through Barton...come on AMD! The EV6 bus can be quad-pumped easily! :confused:

1) GET WET! :D :D
2) They don't want to do up the fsb yet, as the thoroughbred is a yeilds, power and heat move ONLY. Treat it like the same chip as the current XPs, only a newer revision. That's all it is - same pipeline, same cache, same package, same latencies, same ISA's, same basic design. If they up the fsb, then their PR ratings probably do stupid things. The way they've got it ATM, every 0.5 multiplier = 100 more PR. If they upped the fsb, that whole system just wouldn't line up right.

theYANK

10-06-2002, 11:28 PM

Originally posted by KingJackal

1) GET WET! :D :D
2) They don't want to do up the fsb yet, as the thoroughbred is a yeilds, power and heat move ONLY. Treat it like the same chip as the current XPs, only a newer revision. That's all it is - same pipeline, same cache, same package, same latencies, same ISA's, same basic design. If they up the fsb, then their PR ratings probably do stupid things. The way they've got it ATM, every 0.5 multiplier = 100 more PR. If they upped the fsb, that whole system just wouldn't line up right.
You usually make a really good argument, but I can't help with this one...
1. It would be more than logical to up the FSB now, Intel did it with the P3, AMD can do it now, it would breath wonderful new life into a tired(Athlon) line.
2. Upping the FSB wouldn't hurt their PR ratings in my opinion! IT would boost them exponentially! All that bandwidth, and the best core architecture to boot! Then they could REALLY compare themselves to P4's and not the old Tbirds like the current PR ratings "officially" do. Just my $0.02
*EDIT* with JEDEC adopting the PC2700 standard a 166 MHz FSB would really spice things up with all those "DDR333" boards out there :D

KingJackal

11-06-2002, 12:12 AM

It's interesting though. Just how much breathe would the move to 166MHz breath into the Athlons?

The P4's were always thirsty for bandwidth. Their architecture feeds off bandwidth - which is why RDRAM P4 systems are so much faster than their DDR counterparts. When moving to the 133MHz fsb, P4's gained even more bandwidth, and so they understandably started performing like they'd been hyped to be able to!

OK, read this one page and you will see the ( perhaps real ) reason why I don't think AMD will bother:
http://www.anandtech.com/cpu/showdoc.html?i=1595&p=12

Ragnor

11-06-2002, 12:47 AM

The Barton isn't going to be 166mhz fsb either as far as I've read... no 166mhz till hammer

KingJackal

11-06-2002, 12:52 AM

BTW - Anandtech have finally got their ( excellent ) review up:

http://www.anandtech.com/cpu/showdoc.html?i=1635&p=1

Shame about his 2200+ - his 1900+ OC'd further! :rolleyes:

Ragnor

11-06-2002, 01:12 AM

Oh well...

The best thing about the Thoroughbred coming out is that the current Althon XP's are only going to get cheaper..

/continues to watch 1600 - 1800 XP prices

theYANK

11-06-2002, 12:08 PM

Originally posted by KingJackal
OK, read this one page and you will see the ( perhaps real ) reason why I don't think AMD will bother:
http://www.anandtech.com/cpu/showdoc.html?i=1595&p=12
Hmmmmm. I'd never thought that it would have THAT small of an effect on performance! Well I think I definitely need to get off my FSB soapbox, cause I thought Intel was tearing into AMD with all this bandwidth, but now I have a renewed trust in the company I've grown to love. :) Keep cranking out those great cheap chips!!!
*mushy*:rolleyes:

Tiggerz

11-06-2002, 07:02 PM

Originally posted by KingJackal

Are they ALLOWED?
( bearing in mind SSE and SSE2 are proprietary APIs owned and designed by Intel.... :confused: )

Amd license quite a lot of Intel technology. MMX and SSE being two of them.

KingJackal

11-06-2002, 07:23 PM

Originally posted by Tiggerz

Amd license quite a lot of Intel technology. MMX and SSE being two of them.

Mmmmm - but SSE != SSE2.

My point being Intel don't let them use those API's straight away. Otherwise they'd lose an important advantage that their chips always have.... :D

Ragnor

11-06-2002, 07:35 PM

Yup ..

You won't see SSE2 till the Hammer and then intel will have the next thing...

A reasonable analogy is seen in soundcards, ie: Creative' EAX... Hercules and Philips cards use Sensura and Qsound respectively, to emulate EAX v1 and v2 but not EAX HD (as used by the Audigy)... since it's the newest version and that'd have to pay a fat load to Creative for the privillage..

matthew

20-06-2002, 06:31 PM

which is why RDRAM P4 systems are so much faster than their DDR counterparts.

really?...... much faster..? :(

mird-OC

20-06-2002, 08:50 PM

up to 25% faster in bandwidth intensive operation - such as DivX encoding, rendering, etc...

Binky Stunt Cat

20-06-2002, 08:54 PM

almost makes me want a P4 RDRAM system.......almost.....

matthew

21-06-2002, 09:37 AM

o.k.. say that 512mb of RDRAM is faster than 512mb of DDR in these type of bandwidth intensive operations - if it was 1024 of DDR vs 512mb RDRAM, would the RDRAM still be faster (more bandwidth..)

SilverPriest

21-06-2002, 09:43 AM

The quantity of the ram doesnt matter dude.
With RDRAM u NEED to use 2 sticks of it (to my understanding) to get the most out of it.
Read Kingjackals memory guide here (http://www.radiativenz.com/guides/memoryguide/index.shtml).

Wibber

21-06-2002, 09:57 AM

you need 2 sticks period.... I want a tbred daddy, and I want it now!!

mird-OC

21-06-2002, 10:05 AM

like SP said, capacity is different from bandwidth. as an anology, think of a piece of RAM as a bottle - bandwidth would be neck of the bottle. the bottle itself could be a number of capacities (e.g. 750ml, 1l, 1.5l etc).

for example, say you have two bottles - a 4l and a 2l - and you wanted to fill them both with 1l of liquid each. the size of the neck on both bottles is the same, so even tho one bottle has a higher capacity, both will have the same maximum fill (and empty) rate. but if you had another bottle - a 2l with larger neck - you would be able to fill (and empty) that bottle much faster, even tho it has the same capacity as one of the other bottles.

oh and Wibber - with 32-bit RIMMS - no, not anymore :)

Wibber

21-06-2002, 01:28 PM

ah, but can you buy them yet? nope...same as the T-bred, all I'd love that where can I buy it? and none of the actual getting... spot the frustrated one

Deviant

21-06-2002, 05:13 PM

The funny thing with DDRRAM and RDRAM is that DDR to me seem superior (let me explain before the flaming starts)

RDRAM is always in a twin bank configuration, ie you need two sticks to operate. So I think of it as each DIMM share each other clock cycle, and therefore are effictivly working at half spped of the FSB. It isn't however twice the memory bandwidth of DDR at the same speed cause the type of memory has such a high (sh*tty) latency, in fact if it operated at the same speed as DDR it would loose out, that why it runs at 533 MHz etc.

Now DDR can also run in a twin bank configuration, but they don't do it cause current processors mostly AMD would not benifit from it (appart from P4, but that would require new motherboards again for little performance increase again over RDDRAM, remember that intel were locked into RDDRAM for ages by contract). NVIDIA Crush chipset did use DDR in a twinbank DDR configuration, but the on board GeForce cards have to share the doubled bandwidth, so negated any real performance increases, but the NForch does keep up with KT266a.

Will we see Hammer with twin bank technology? Maybe? I dunno. Maybe I'm missing something and someone can shed some light.

mird-OC

21-06-2002, 07:33 PM

what you're saying is basically true, but the only thing that DDR has over RDRAM is it's latency. there are hundreds of articles out there explaining why RDRAM is and always will be (by design)superior to DDR (and DDR-II).

Deviant

21-06-2002, 09:17 PM

DDR II is just waiting until the time is right, technology is there but not the time. I think in a few years RDDRAM will be the norm when DDR can not longer take us all the way, and RDDRAM will be more mature, but DDR has alot more to prove yet.