Actually, that's what I was basing my statement of the previous discussions about the Hammer and bandwidth. If I remember correctly, there was a point raised about pairing RDRAM with the Athlon, but then there was raised the issue of the Athlon actually having the architechre (I know, my spelling sucks) to take in all that info that RDRAM could push through. Then, when the possibility of the Hammer having a 400MHz FSB was brought up, again, the idea of pairing it with RDRAM came up, the thinking that with such a large FSB, it must be able to handle copious amounts of data. So, since there is no seeming placement of RDRAM on the next year's roadmap, does that mean that the Hammer isn't going to need those large amounts of bandwidth to function at full potential? Or does it mean that AMD sees no future for RDRAM? Or is there another reason? And, if Raystonn and others in the RDRAM business are to be believed, the latency will decrease as the MHz of the memory is increased. Just like the P4, it seems that RDRAM wasn't really meant for low speeds, and is suffering because of it.
-SammyBoy
-SammyBoy
- Sep 3, 2001
- 468
- 0
- 18,780
I'd rather not talk about memory bandwidth..you can go and visit Raystonn's post for that. Remember Hammer will have an integrated on chip memory controller and therefore memory latencies will be drastically reduced even at low clock speeds.
I would like to include this extract from Anandtech...
<font color=blue>AMD could theoretically produce a Hammer with DDR SDRAM support and another with RDRAM support just by changing the DCT, but to end all speculation now, RDRAM would make very little sense on the Hammer. Remember that one of the downsides of RDRAM is increased latency in many situations; one way of hiding this latency is by pairing RDRAM with deeply pipelined CPUs such as the Pentium 4. It's obvious by now that the Hammer isn't as deeply pipelined of a CPU and won't have the clock speed to offset RDRAM latencies as well as the Pentium 4. This also makes AMD's decision to continue to support DDR SDRAM very sensible.</font color=blue>
<A HREF="http://www.anandtech.com/cpu/showdoc.html?i=1546&p=6" target="_new">Source</A>
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
I would like to include this extract from Anandtech...
<font color=blue>AMD could theoretically produce a Hammer with DDR SDRAM support and another with RDRAM support just by changing the DCT, but to end all speculation now, RDRAM would make very little sense on the Hammer. Remember that one of the downsides of RDRAM is increased latency in many situations; one way of hiding this latency is by pairing RDRAM with deeply pipelined CPUs such as the Pentium 4. It's obvious by now that the Hammer isn't as deeply pipelined of a CPU and won't have the clock speed to offset RDRAM latencies as well as the Pentium 4. This also makes AMD's decision to continue to support DDR SDRAM very sensible.</font color=blue>
<A HREF="http://www.anandtech.com/cpu/showdoc.html?i=1546&p=6" target="_new">Source</A>
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
- Sep 3, 2001
- 468
- 0
- 18,780
So what do you think the actual clock speed of the PR 3400 Hammer is ?? The CNET artical says that my simply subtracting 300-400 off the PR value like the Athlon won't reveal the clockspeed, in fact he says the diffrence is greater.
I think the Hammer PR 3400 might be running at around 2.5Ghz.
P.S A PR 3400 Hammer might more likley be equvilant to a P4 3.6GHz if AMD contiune to be modest in thier ratings.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
I think the Hammer PR 3400 might be running at around 2.5Ghz.
P.S A PR 3400 Hammer might more likley be equvilant to a P4 3.6GHz if AMD contiune to be modest in thier ratings.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
- Sep 3, 2001
- 468
- 0
- 18,780
Even with Jackson SMP P4 might have a hard time with Hammer. Am I right in thinking Jackson offers a 30% increase in performance therefore a Jackson enabled P4 @ 3Ghz will be equivalent to a normal P4 @ 3.3-3.4Ghz ???
With a PR 3400 this would Hammer in the lead if not on par with P4 Jackson @ 3Ghz which is what Intel are planning in 3Q02.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
With a PR 3400 this would Hammer in the lead if not on par with P4 Jackson @ 3Ghz which is what Intel are planning in 3Q02.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
- Sep 3, 2001
- 468
- 0
- 18,780
That would still make Hammer on par with P4 Jackson, assuming the PR values aren't modest.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
lhgpoobaa
Illustrious
- Dec 31, 2007
- 14,456
- 1
- 40,780
i was just thinking (yes i do think from time to time)....
64mm squared die size...
thats 8mm x 8mm...
thats one friggin tiny chip!
they will be able to fit a crapload onto a 200 or 300 mm wafer!
OEMs selling "High End"PCs with integrated video will be forced into Q3tournaments using a TNT2M64!
64mm squared die size...
thats 8mm x 8mm...
thats one friggin tiny chip!
they will be able to fit a crapload onto a 200 or 300 mm wafer!
OEMs selling "High End"PCs with integrated video will be forced into Q3tournaments using a TNT2M64!
Parma_Endorion
Distinguished
- Nov 10, 2001
- 34
- 0
- 18,530
I don't think IA64 will ever come to the desktop. IA64 is a failure. During the last decade Intel has been defending its CISC x86 against several other RISC challengers (PPC, Alpha...) with a lot of FUD. When it was time to extend the ISA to 64 bits, Intel had two options. Extend x86 to 64 bits or create a new 64-bit RISC ISA. One was technically challenging and the other was very embarrassing.
Quote from RWT:
"HP came along and offered a third alternative. HP told Intel a great story about how it had cured VLIW of its inherent problems by adding new architectural features, which also preserved its basic simplicity of control logic. This was kind of like getting "all the great taste of OOO execution superscalar performance but without the control logic calories". Since the term VLIW was beginning to have a bad odour associated with it Intel and HP had to come up with a new term to replace it. Thus EPIC was born."
And another one...
"But face it, the native performance of the first two generations of IA64 processors show hardly any performance differentiation over native x6 processors. Intel and HP chose an obscure, niche oriented architecture style - VLIW. The only redeeming quality of VLIW is it allows parallel instruction execution with minimal control logic. But there is no free lunch however and the downside of VLIW is that it performs rather poorly when running code not dominated by processing vector type data within loops. Unsurprisingly, the only success that VLIW has found is in the field of DSPs. The first VLIW machines were mid-range machines for technical computing but VLIW's drawbacks were so severe that it couldn't survive in even this narrow niche."
Quote from RWT:
"HP came along and offered a third alternative. HP told Intel a great story about how it had cured VLIW of its inherent problems by adding new architectural features, which also preserved its basic simplicity of control logic. This was kind of like getting "all the great taste of OOO execution superscalar performance but without the control logic calories". Since the term VLIW was beginning to have a bad odour associated with it Intel and HP had to come up with a new term to replace it. Thus EPIC was born."
And another one...
"But face it, the native performance of the first two generations of IA64 processors show hardly any performance differentiation over native x6 processors. Intel and HP chose an obscure, niche oriented architecture style - VLIW. The only redeeming quality of VLIW is it allows parallel instruction execution with minimal control logic. But there is no free lunch however and the downside of VLIW is that it performs rather poorly when running code not dominated by processing vector type data within loops. Unsurprisingly, the only success that VLIW has found is in the field of DSPs. The first VLIW machines were mid-range machines for technical computing but VLIW's drawbacks were so severe that it couldn't survive in even this narrow niche."
Parma_Endorion
Distinguished
- Nov 10, 2001
- 34
- 0
- 18,530
Jackson SMT will only give a maximum of 30% higher performance and only if the application is written for multiprocessing. In most cases Jackson SMT will NOT give any advantage at all.
Then there is also another black cloud called AMD. Amd’s dual core hammer will give a higher performance increase, compared to Jackson SMT, since the AMD approach also doubles the execution units.
Then there is also another black cloud called AMD. Amd’s dual core hammer will give a higher performance increase, compared to Jackson SMT, since the AMD approach also doubles the execution units.
- Sep 3, 2001
- 468
- 0
- 18,780
True..all those 32bit apps will have to be recompiled to take advantage of Jackson.
I think the most important thing to remember is that AMD are bringing 64 bit computing to the DESKTOP and LAPTOP i.e the common user. This is quite an achievement. Whether common users need this is another question.
Sure its 64 bit performance my not be as fast as Itainum but I am sure it is 'reasonable'. This is better than nothing. We will just have to wait till the benchmarks for Hammer vs Itainum come out.
If there is a large difference in 64 bit performance it would mean Hammer will not be suited to the 64 bit server market.
However companies may be willing to a compromise performance if the price of Hammer is considerablly lower, which is probably what will happen.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
I think the most important thing to remember is that AMD are bringing 64 bit computing to the DESKTOP and LAPTOP i.e the common user. This is quite an achievement. Whether common users need this is another question.
Sure its 64 bit performance my not be as fast as Itainum but I am sure it is 'reasonable'. This is better than nothing. We will just have to wait till the benchmarks for Hammer vs Itainum come out.
If there is a large difference in 64 bit performance it would mean Hammer will not be suited to the 64 bit server market.
However companies may be willing to a compromise performance if the price of Hammer is considerablly lower, which is probably what will happen.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
- Sep 3, 2001
- 468
- 0
- 18,780
Yes PR ratings are based on the T-bird but surely you can use this rating to compare it to a Northwood or is this too simplistic ? Let me explain.
Athlon XP1800 = 2Ghz P4....a P4 NorthWood @ 2Ghz will out perform AXP1800/1900 therefore I assume:
Athlon XP 2000 = 2Ghz Northwood P4
(I know this is based on assumptions. I hope you understand my logic, correct me if I am wrong).
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
Athlon XP1800 = 2Ghz P4....a P4 NorthWood @ 2Ghz will out perform AXP1800/1900 therefore I assume:
Athlon XP 2000 = 2Ghz Northwood P4
(I know this is based on assumptions. I hope you understand my logic, correct me if I am wrong).
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
Yummy - Sep 3, 2001
- 468
- 0
- 18,780
Out of curiosty how big is IA64 ? Are there any websites that have a picture of it and technical specs e.g core voltage etc. ?
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
Parma_Endorion
Distinguished
- Nov 10, 2001
- 34
- 0
- 18,530
- Sep 3, 2001
- 468
- 0
- 18,780
Hmmm...now that would be interesting ! I would like to see some benchmarks for that ! However maybe the two technologies are mutually exclusive either you have SMP optimised or Jackson optimised not both. Anyone confirm this ?
That would really push the P4/Xeon Jackson...most likely it would overheat. Sorry I meant cause the system to shut down (forgot about P4's amazing throttling capability).
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
That would really push the P4/Xeon Jackson...most likely it would overheat. Sorry I meant cause the system to shut down (forgot about P4's amazing throttling capability).
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
- Sep 3, 2001
- 468
- 0
- 18,780
..."flaming crap". Hee Hee 
excuse the pun.
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
excuse the pun.<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
Parma_Endorion
Distinguished
- Nov 10, 2001
- 34
- 0
- 18,530
The same code will run for SMT & SMP just as it does for Dual CPU's today.
/* The more you know, the more you realize how little you know */
/* The more you know, the more you realize how little you know */
flamethrower205
Illustrious
- Jun 26, 2001
- 13,103
- 0
- 40,780
What if teh Hammer's PR rating is based on the latest Athlon then. Imagine if the Thoroughbred is based on Pally for PR, and is 33% faster, and then the Hammer. Cool. Of course I'm probably wrong, and just looking for something to drool over
If it's working...overclock it!
If it's working...overclock it!
- Sep 3, 2001
- 468
- 0
- 18,780
Seems like AMD are going to agressively increase performance with the Hammer. Do you think AMD will do a straight jump from 3400 to 4000 or produce smaller increments like 3500..3600.
Either way with a 4400 in 2H/03 it looks like Hammer has a lot of potential performance wise. If AMD keep with this schedule and PR rating it would mean Intel will have to increase the clock rate of the P4 by 1.5Ghz a year just to keep up.
<font color=blue>The Hammer 3400 will be followed by the Hammer 4000 during the first half of 2003 and the 4400 in the latter part the year, when the Hammer family is shrunk using as 0.09 micron (though some AMD documentation says "0.10 micron") production process. Hammer will debut at 0.13 micron.
</font color=blue>
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
Either way with a 4400 in 2H/03 it looks like Hammer has a lot of potential performance wise. If AMD keep with this schedule and PR rating it would mean Intel will have to increase the clock rate of the P4 by 1.5Ghz a year just to keep up.
<font color=blue>The Hammer 3400 will be followed by the Hammer 4000 during the first half of 2003 and the 4400 in the latter part the year, when the Hammer family is shrunk using as 0.09 micron (though some AMD documentation says "0.10 micron") production process. Hammer will debut at 0.13 micron.
</font color=blue>
<font color=purple>~* K6-2 @ 333MHz *~
I don't need a 'Gigahertz' chip to surf the web just yet ;-)</font color=purple>
TRENDING THREADS
-
-
-
-
Discussion What's your favourite video game you've been playing?- Started by amdfangirl
- Replies: 4K
-
-
Question Home phone and pc compromised by malicious person
- Started by geoffranklyn
- Replies: 4
Facebook
Twitter
Instagram