The Thouroughbred and the Hammer

Quetzacoatl

Distinguished
Jan 30, 2002
1,790
0
19,780
Sup guys. Okay, got some questions and some speculation. As many of us know, sometime soon (anyone got any ideas), AMD will be releasing it's new Thouroughbred processor, which begs the big question, will it still be a K-7 and be supported by SocketA? Also, if it's a new "K-8", what kind of chipset will we be looking forward to? As for later on this year (I think about late Q3), AMD plan's on releasing the so called K-8 Hammer. Now, according to rumors, it's gonna smash anything Intel has, namely the well known PentiumIV northwood. What can we expect of Hammer? Will it have more cache, a faster FSB, perhaps SSE2 or other extensions? I mene, if AMD plans on beating out Intel really badly, they need to take a step beyond the Athlon XP and really lay it on Intel.

Now, I invested a little bit of $$$ into AMD in stocks, and i'm assuming the release of the Thouroughbred and the Hammer will send AMD skyrocketing in the stock market IF they advertise enough. I'd be willing to bet that commecials and graphs would persuade people to move away form Intel. Does anyone think I should ditch or keep my AMD stock. Right now, i'm geared towards holding to it until Thouroughbred, and perhaps even Hammer. Now, i'm not biased against Intel, but seriously, if they really boast to have the fastest X86 processor anywhere, you'd think they could really run away with 2.2Ghz and a 400mhz QDR FSB. They are barely beating AMD, and only under some circumstances. Heck, thats like a 500 Mhz lead, and AMD can beat Intel in some benchmarks (AXP1.67 vs PentiumIV2.2). Well, if anyone has any insight, i'm all ears.

"When there's a will, there's a way."
 

FatBurger

Illustrious
First, you could probably learn more by searching for relavant threads, but I'm happy to answer your questions for now.

Thoroughbred is a die shrink. They're moving from .18 micron to .13 micron. There are rumors that it will be a 166 (333) bus, but they are very unconfirmed. Still socket A, but probably not all motherboards will support it.

Hammer is completely different technology. New socket, new assembly language, etc. SSE2 is the only addition that AMD has spoken of so far.

<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>
 

Matisaro

Splendid
Mar 23, 2001
6,737
0
25,780
According to the register it is still 133fsb, and any motherboard which works with the axp will work with tbred.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
 

lhgpoobaa

Illustrious
Dec 31, 2007
14,462
1
40,780
well i dont expect 166fsb until the start of SIO & barton... cant see it on any 'near future' roadmaps.

and it will be a bit of a shame to see socket A finally go. ahh well.

MasterPoo@mybeast.myhome.melbourne.vic.au.sothernhemisphere.3rdplanet.sol.milkyway.universe
 

eden

Champion
Hey, at least it ain't no 423/478 fiasco and constant socket changes like Mr. Intel makes us do!

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...
 

eden

Champion
Anandtech may had rumors on it but it's a very respectable website and I believe most of what they claimed about the Hammer.
Check there for info but here's what I know:
-It will definitly have 512K cache for home users, the ClawHammer.
-Memory Controller on CPU, meaning huge IPC.
-Improved pipeline by 12 stages now, a 20% increase. Better than making like Intel and put 20 stages, ruining IPC completly. It will allow better speeds for AMD by a tad but it won't ruin IPC as they have so many measures to counter the pipeline increase.
-0.13m which also means more power.
-64-bit and 32-bit support means that we will finally be able to make a comfortable but leveled transition without hassles.
-SSE2 is a good target for AMD to develop as its processors are right on the power to handle such, and it will prove more useful than on the P4.
- FPU and ALU-wise I have not heard of any changes but it's the memory controllers that will make do for a huge jump in performance.

It is rumoured to begin at PR 3400+, at 2GHZ. Whether this is true or not it brings confusion at some point. For example is it in comparison to the TB or P4 or AXP now? Would that mean that it performs as good or better than a P4 NW 3.8GHZ? If that is the case then we have arrived at an era of ultimate computing, thanks to AMD as 2GHZ performing such power is beyond what Macs G4s had, and will truly show Intel how hard word pays off. We are indeed witnessing disciples of technology bringing us a true power. I do hope these are all true specs and that they will bring forth new generation of computing. I doubt it will be anywhere lower performing, because anything from AMD right now, is nothing but even higher IPC as if it wasn't enough now! And as we stand, the Tbred will bring XP to high speeds which will take back the lost sentence: "More MHZ, more performance" which Intel destroyed with their P4.... And then if that isn't enough the Claw Hammer will just put even more cream on the banana and add a cherry on top, to truly be the next generation....I HOPE!



These are speculations and you can't say anything against me, because it's from them and to be honest, it makes sense.

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...
 

texas_techie

Distinguished
Oct 12, 2001
466
0
18,780
Quetzacoatl, is your name a reference to the Spanish promised land? I seem to rememeber something about that in my philosophy course.

Anyway, if you do a search on google for "amd hammer" you will find numerous technical articles on Hammer. Yesterday I read about 15 article in that one search alone. Second, AMD will not advertise a whole lot, mainly because they never have. They dont have the budget for big advertising campaigns.
WIll AMD stock go up because of Hammer? Probably, but Id stay with the stock until at least Q2 2003. The chip wont be out until q4 this year at the earliest. So you need to give it time to saturate the market.
The k-8 chipset is a complete mystery. I havnt seen a single article on the chipset itself. I know a guy who is in the chipset design dept. for AMD, but he isnt saying a word. The roadmaps of SIS, ALI and Via give a slight clue. Like USB2, Hypertrans etc. But how the chipset will perform is unknown. The AMD refernce tends to be slow, but very reliable. OEMs and mobo makers take the reference design and add to it. Often they use inferior parts though, degrading the chipsets quality. Anyway, If anyone has heard about the reference design for Hammer chipsets feel free to chime in here.

Benchmarks are like sex, everybody loves doing it, everybody thinks they are good at it.
 

leonov

Distinguished
Dec 31, 2007
66
0
18,630
http://www.dj.st44.arena.ne.jp/xwin2/mainhtml/intelcpu/amdcodename2002.html

If you look here you can see some interesting things. The Hammer is introduced at M3400 in Q402 and the following quarter we see M4400. Why such a huge jump? It *may* be that they will introduce a Socket A version in Q4 and have the on die memory controller version for Q103.

Also note Barton has been moved back to 2003, perhaps because they don't want to try to intro it when they are looking at getting the the Hammer ready.

L
 

tlaughrey

Distinguished
May 9, 2001
581
0
18,980
Quetzacoatl, is your name a reference to the Spanish promised land? I seem to rememeber something about that in my philosophy course.

Quetzalcoatl
Pronounced As: ketsälkôätl [Nahuatl,=feathered serpent], ancient deity and legendary ruler of the Toltec in Mexico. The name is also that of a Toltec ruler, who is credited with the discovery of corn, the arts, science, and the calendar. It is unclear whether the ruler took his name from the god or as a great ruler was revered and later deified.

<i>I made you look. But I can't make you see.</i>
 

bikeman

Distinguished
Jan 16, 2002
233
0
18,680
And as we stand, the Tbred will bring XP to high speeds which will take back the lost sentence: "More MHZ, more performance" which Intel destroyed with their P4....
Aaarghhh!!! Why is everybody nagging this much about this IPC (which, I assume, stands for Instructions Per Clock cycle?)? Isn't it the benchmark and real time performance that matters? You say it yourself: Macs have had a much bigger performance to clock speed ratio for months or years, and was everybody complaining to AMD and Intel for not having the same ratio? I don't think so. Intel just chose to take the high-clock-speed approach, left aside wether their goals were technical or commercial. The results are clear, though. AMD runs at 1.67 GHz, Intel at 2.2 GHz and they perform on par on average. Intel and AMD just chose to go another way to reach the same point I guess. And who knows, maybe Intel will be capable of returning to the same IPC than they reached on the PIII-series by finetuning their architecture (although the 20-stage pipeline makes that quite hard ...). Who knows ... We all will know, but we just need to be patient for what the future brings.

Hmmm ... I appear not to be quite good at writing short posts. All this text just to kindly ask everybody to stop complaining about this IPC-stuff ... I'm sorry ...
 

eden

Champion
Simply put: Intel and AMD are PCs. Macs are MACS. Each is a different thing and is not made to be compared at all.
Until AMD goes to Mac world, the IPC comparison is the most valid way of seeing what PC is the true one.

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...
 

Matisaro

Splendid
Mar 23, 2001
6,737
0
25,780
And who knows, maybe Intel will be capable of returning to the same IPC than they reached on the PIII-series by finetuning their architecture


Thats impossible, to have high clockspeed you MUST have lower ipc, intel made a trade off, no coming back.

"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
 

eden

Champion
They can if they do like AMD for having increased 20% pipeline, they added memory controlers and other stuff to compensate and increase IPC. It is possible, especially once they reduce to 0.09m they should be able to add the FPU back and increased L1 cache.

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...
 

Matisaro

Splendid
Mar 23, 2001
6,737
0
25,780
They can if they do like AMD for having increased 20% pipeline, they added memory controlers and other stuff to compensate and increase IPC. It is possible, especially once they reduce to 0.09m they should be able to add the FPU back and increased L1 cache.

Yes, but your forgetting that theipc takes the clock speed total and thats part of the final number, meaning a 3.5ghz cpu which performs the same as a 2ghz cpu neccicarily has lower ipc than the 2ghz cpu. Because it has a higher clock, and the same performance.

Now, the p4 has many stages to increase clock, and even if they added functionality to give better performance per clock, as long as their goal is to have a high clock speed, they must always have a lower ipc. Follow me?




"The Cash Left In My Pocket,The BEST Benchmark"
No Overclock+stock hsf=GOOD!
 

eden

Champion
Ah yes I get that, so in a way, Intel really has shun us from the performance society and has put itself with dumbnuts that think big numbers means big power... how sad of them.


PS:
This is off-topic but I needed quick help here: I have an LG 16*10*40 CDRW that has read problems: When playing some games it would stop spinning instead of going on, and that causes games to hang waiting for it respin, and in any game where there is track change or music change, it has to also stop and respin, which causes 5-6 seconds of wait and that can make any online game go crap. I've contacted so many places but none helped me and only said to check IDE controllers. Should I install the new VIA 4-in-1 drivers for WinXP? Or should I let WinXP as it is with the stock 4 in 1? And what are my risks of doing this update? I am worried it would cause conflicts or may not even help in WinXP. Other than that what would you suggest???

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...
 

SammyBoy

Distinguished
Mar 16, 2001
689
0
18,980
<A HREF="http://freespace.virgin.net/m.warner/RoadmapQ402.htm" target="_new">Here's</A> a link that has info on the AMD Hammer chipsets. The info is sparse, but most info about AMD products is until the actual release date. They seem to enforce NDAs well enough that people don't dare cross them. Also, on the previous quarter there is info on ALi and SiS Hammer chipsets.

-SammyBoy
 

eden

Champion
Hmm Windows Longhorn, it had better be a codename!
As for the ClawHammer, that's pretty much what I had covered so it should indeed have such powerful specs. In any case a processor able to defeat 1GHZ+ gaps of difference is mighty indeed.

Please could anybody help me btw? Read my previous post and help!!

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...
 

IIB

Distinguished
Dec 2, 2001
417
0
18,780
Intel didnt "destroy" IPC.

the P4 has lower IPC and Higher Clock speeds
its just as good as
higher IPC and lower clock speeds (athlon)

at the end of the day you compare the preformance of Intels latest offering with AMDs...

if P4 sucks then Athlon must suck as well... and thats the bottom line.


This post is best viewed with common sense enabled
 

FatBurger

Illustrious
the P4 has lower IPC and Higher Clock speeds
its just as good as
higher IPC and lower clock speeds (athlon)

As long as they balance out, of course. No use in having 1 IPC and 1GHz instead of 5 IPC and 800MHz :wink:

<font color=orange>Quarter</font color=orange> <font color=blue>Pounder</font color=blue> <font color=orange>Inside</font color=orange>
 

zengeos

Distinguished
Jul 3, 2001
921
0
18,980
"http://www.dj.st44.arena.ne.jp/xwin2/mainhtml/intelcpu/amdcodename2002.html

If you look here you can see some interesting things. The Hammer is introduced at M3400 in Q402 and the following quarter we see M4400. Why such a huge jump? It *may* be that they will introduce a Socket A version in Q4 and have the on die memory controller version for Q103."

It's not at all uncommon for early versions of chips to have relatively low speeds compared to just a few months later. Take for example the Palomino debu at 1-1.2ghz in MP and Athlon 4 form. AMD couldn't produce enough high speed chips early on. So, they produced lower speed chips and marketed them as mobile and MP chips. Smart move AND it allowed them to work on their yields AND increase speeds. My suspicion is AMD is expecting some bumps early on in production, so will produce chips at an estimated PR3400, with production ironing out in the following 6 months to allow for a substantial speed boost. Hammer will NOT be a Socket A chip. It makes no sense for AMD to produce a SocketA version.


"Also note Barton has been moved back to 2003, perhaps because they don't want to try to intro it when they are looking at getting the the Hammer ready.
"

It has? Last time I looked (5 minutes ago) at AMD's roadmap, it listed Barton as being released 2H02. So, I don't know where you get the idea it's been puhed back to 2003. Of course, AMD may have just neglected to update their roadmaps to reflect the change....


Mark-

When all else fails, throw your computer out the window!!!
 

IIB

Distinguished
Dec 2, 2001
417
0
18,780
true
for now - they seems to blance out.
as no company had sgnificant lead in preformance for alot of time now.


This post is best viewed with common sense enabled
 

eden

Champion
But when the AXP with Tbred goes up to NW speeds, then it will paint a very different picture as IPC and MHZ will play a very powerful role for Athlons then... but that's to be seen very soon, can't wait for March!

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...
 

IIB

Distinguished
Dec 2, 2001
417
0
18,780
there will still be a Mhz gap
when T-bred hit say 2Ghz intel will already have 2.5-2.6Ghz P4 to comabat it...

the P4 can reach Higher clock speeds - its a given.

Intel and AMDs latest processors seem to even out - so no processor is much better then the other.
its diffrente approch:

one takes the higher clock speed lower IPC path
the other the Higher IPC lower clock speed way

in the end it seems that no company can take the absolute lead in what counts - Preformance.

it evens out...

This post is best viewed with common sense enabled