Prescott have the potential of the A64?

[LordSnailz]

Originally posted by: dguy6789

Originally posted by: SickBeast

Originally posted by: dullard
Here are my thoughts.

1) Prescot should come out at 3.4 GHz - giving a 5% boost over the current top 3.2 GHz P4.
2) Prescot will have more cache, improved hyperthreading, new commands, etc. People claim it will have a 15% boost to efficiency - I'll use 10% here to be conservative.
3) The net effect of (1) and (2) will be roughly a 15% boost from the current 3.2 GHz P4.
4) Workstation wise, the P4 runs just as well as the Athlon 64, but gaming wise the P4 needs help. Here is a good starting point for the Athlon 64 benchmarks. This is finally an article without overclocking the Opteron.

I'll focus on gaming since that is what most people here want fast chips for. Looking at that review, the P4 needs a good 15%-20% boost to be competitive. That is exactly what (3) says it will get. So I think we will see another neck and neck battle. It just depends on who can ramp up clock speed faster.

Interesting. You're neglecting to note that the Athlon FX is rumoured to perform 40-50% faster than its P4 brethren in some situations tho.

The other thing that you're not keeping in mind is the fact that the FX can do 64-bit.

An integrated dual-channel memory controller is icing on the cake. Hyperthreading interests me, but computers have been able to do multitasking for years now. The one thing that intel will have in its favour is the 0.09um fab process. The next few months should be a nice heated battle to watch.

the athlon 64 FX is 0.09nm as well, intel does not have an advantage in size.

link? I was under the impression that intel was the only fab having some sucess with the .09nm process ...

 

[batmang]

if my rig with a top of the line video card doesnt run half life 2 flawlessly, im giving up on future gaming investments. seriously. i think we, the consumers, spend way to much fvcking money UPGRADING to play games, then when they come out they own your systems, and your forced to upgrade even more. i really hope these up and coming games are actually fun, instead of just eye candy, from what i heard at quakecon about doom 3, its just eye candy and the gameplay sux0rs.

 

[INemtsev]

Of course half life will play with A64 duh...cmon...put some belief in that newcoming processor.....

 

[Evdawg]

1) Assume the 3.06 GHz P4 gets 100 on a benchmark.
2) Assume the Athlon 64 is 40% to 50% faster - thus it gets 140 to 150 on that benchmark.
3) Add 10% to the 3.06 GHz P4 to see what the 3.2 GHz P4 gets: 100*1.10 = 110.
4) Add 5% for movement from 3.2 GHz to 3.4 GHz: 110 * 1.05 = 115.5.
5) Assume the Prescot improvements give it a 10% conservative boost over the current P4: 115.5 * 1.1 = 127.1.
6) Thus the Athlon 64 is 140/127.1 - 1 = 10% faster to 150/127.1 = 18% faster. That is damn impressive. 10% to 18% faster. But again that was on SOME programs.

Wow dullard im impressed by your math skillz . Wow .09 micron processors.... just when you thought the coolers couldnt get any bigger, and it gets SO hot, they make a .09, thats pretty impressive. I used to run a 1.4 athalon with a .15, it ran 130 idle. Now ive oc'd my xp2500 to 2.2 ghz and i run 95 idle.

Personally, i have a feeling that the a64 is gonna have a good lead over the prescot. All i see a diff in between the current p4c's and the upcoming presscots are the cache, and clock speeds. The Xp3000 and P4b 3.06 have a semi close relationship. Its not real close but its not HUGE. The clock speed of a xp 3000 is 2.1, the p4 3.06 is 3006 ghz. Considering theres almost an entire ghz of clock speed gap, AMD does a helluva job with such a low clock speed and fsb. Thats what i look at... and value of course

 

[Duvie]

Originally posted by: Evdawg

1) Assume the 3.06 GHz P4 gets 100 on a benchmark.
2) Assume the Athlon 64 is 40% to 50% faster - thus it gets 140 to 150 on that benchmark.
3) Add 10% to the 3.06 GHz P4 to see what the 3.2 GHz P4 gets: 100*1.10 = 110.
4) Add 5% for movement from 3.2 GHz to 3.4 GHz: 110 * 1.05 = 115.5.
5) Assume the Prescot improvements give it a 10% conservative boost over the current P4: 115.5 * 1.1 = 127.1.
6) Thus the Athlon 64 is 140/127.1 - 1 = 10% faster to 150/127.1 = 18% faster. That is damn impressive. 10% to 18% faster. But again that was on SOME programs.

Wow dullard im impressed by your math skillz . Wow .09 micron processors.... just when you thought the coolers couldnt get any bigger, and it gets SO hot, they make a .09, thats pretty impressive. I used to run a 1.4 athalon with a .15, it ran 130 idle. Now ive oc'd my xp2500 to 2.2 ghz and i run 95 idle.

Personally, i have a feeling that the a64 is gonna have a good lead over the prescot. All i see a diff in between the current p4c's and the upcoming presscots are the cache, and clock speeds. The Xp3000 and P4b 3.06 have a semi close relationship. Its not real close but its not HUGE. The clock speed of a xp 3000 is 2.1, the p4 3.06 is 3006 ghz. Considering theres almost an entire ghz of clock speed gap, AMD does a helluva job with such a low clock speed and fsb. Thats what i look at... and value of course

Yep, but that is larger L1 and L2 cache...Also remember there will be 13 new added instruction sets which will enhance thjings as well....Then there is rumored to be a better rework of HT which may help considerably. Remember HT in the Xeons versus HT in the p4c's showed that Intel had already refined it once with good success...

I am thinking the prescott in the longer terms in not an answer and a good comparison to the A64. Intel however has already stated they saw no need for 64bit computing in the destop home user market. When they do we will see more of an answer to the A64. I think the prescotts pure mhz speed and larger cache and what not will keep it close for awhile especially if Intel can at least get 4ghz out of the prescott. That may constitute to maybe a 4.5ghz+ northwood if Intel Pr rated!!!

It all really depends on how big a lead AMD gets and how fast this chip will ramp up. The heat seems good so I think amd can ramp up well. Financial factors and market factors may dictate timing of releases....

 

[Jeff7181]

Intel however has already stated they saw no need for 64bit computing in the destop home user market.

There was a time when people thought there was no need for computers in homes period.

Just because Intel doesn't see a need for it, doesn't mean there's a purpose for it. If you can do more work at once time with less strain, that's a good thing, right? I think it is. And if you can do more work with less strain, you can do A LOT more work with the same amount of strain. 64-bit computing is undoubtedly faster than 32-bit computing Mhz for Mhz... so the question to me is... why NOT move to 64 bit???

*EDIT* If Intel doesn't think it's necessary right now, when will it be necessary in their eyes?

Intel doesn't see a need for it because current software just doesn't need that kind of computing power... well... DOS didn't need a 133 Mhz Pentium processor to run well either...

My point is, faster, more efficient hardware will pave the way for better software. In the past it has been the other way around, and I'm glad to see that's changed... with my first 386, it seemed like it took 5 minutes for Windows 3.1 to boot up. I could start MS Word, go into the kitchen, get a glass of water, grab a couple cookies, and come back to the computer and finish waiting for MS Word to load. That's unheard of in today's world of computers... and I'm glad... I don't want to have to wait for MS Word to load, when I click it, I want it up there and ready to use instantly. As things get more complicated, new file systems, more intuitive software, more personalized software... we're going to need even faster hardware. And quite frankly, I don't want to have to wait for the hardware to use new software.

 

[Duvie]

Originally posted by: Jeff7181

Intel however has already stated they saw no need for 64bit computing in the destop home user market.

There was a time when people thought there was no need for computers in homes period.

Just because Intel doesn't see a need for it, doesn't mean there's a purpose for it. If you can do more work at once time with less strain, that's a good thing, right? I think it is. And if you can do more work with less strain, you can do A LOT more work with the same amount of strain. 64-bit computing is undoubtedly faster than 32-bit computing Mhz for Mhz... so the question to me is... why NOT move to 64 bit???

I am not arguing or agreeing with INtels claim...I am merely stating their interpretation of the state of pc home user market....They stated this many times before and I don't think they have changed their minds yet. However AMD and its success and success in getting OSes designed for it may give them a kick to get going....

 

[LordSnailz]

Originally posted by: Jeff7181

Intel however has already stated they saw no need for 64bit computing in the destop home user market.

There was a time when people thought there was no need for computers in homes period.

Just because Intel doesn't see a need for it, doesn't mean there's a purpose for it. If you can do more work at once time with less strain, that's a good thing, right? I think it is. And if you can do more work with less strain, you can do A LOT more work with the same amount of strain. 64-bit computing is undoubtedly faster than 32-bit computing Mhz for Mhz... so the question to me is... why NOT move to 64 bit???

*EDIT* If Intel doesn't think it's necessary right now, when will it be necessary in their eyes?

Intel doesn't see a need for it because current software just doesn't need that kind of computing power... well... DOS didn't need a 133 Mhz Pentium processor to run well either...

My point is, faster, more efficient hardware will pave the way for better software. In the past it has been the other way around, and I'm glad to see that's changed... with my first 386, it seemed like it took 5 minutes for Windows 3.1 to boot up. I could start MS Word, go into the kitchen, get a glass of water, grab a couple cookies, and come back to the computer and finish waiting for MS Word to load. That's unheard of in today's world of computers... and I'm glad... I don't want to have to wait for MS Word to load, when I click it, I want it up there and ready to use instantly. As things get more complicated, new file systems, more intuitive software, more personalized software... we're going to need even faster hardware. And quite frankly, I don't want to have to wait for the hardware to use new software.

you make a lot of valid points but to answer you "why not move to 64bit", you answered it yourself with intel's statement. There is no need. You have to understand that a majority of computer users are not gamers but your normal browsing the web, office app. user. In your everday apps. the 64bit processor does not mean you will do more compared to your 32bit processor. Why spend all the R&D money into developing a 64bit processor for the home desktop when you can improve and scale your current process?

I don't think your dos+133Mhz pentinum was a good example ... if you take the highest clocked cpus in the market, a majority of the people, do not need that fast of a computer. This is evident from the lack of sales from the past year. People are not upgragding their computers as often because they don't see a need in it. Yes, I think the hardware is ahead of the software people right now. Having said that there's always that niche of games, overclockers, and engineers that would benefit from the fastest computer out there or from a 64 bit processor. But the real money is in your typical computer user.

 

[ForsakenMaster]

Personaly i do not beleive you can get a acurate comparison of the 2 CPU 's unless it was possible to put them on the same MOBO.
As the past has showed us P4 v AMD comparisions have almost always been biased to intels favor,plus we should take this fact into account,what software will be able to untilize the FULL potential of a 64 bit CPU ?
None at the moment i am aware of.

 

[jonny13]

I think Intel will want to move to a 64-bit desktop sometime, I just don't think they want to use AMD's approach, and not just because AMD is doing it first. For one, AMD is just adding on 64bitness to the 25 year old x86 language. I think Intel (and others) want to move to something more efficient and updated. When I look at the Itaniums, I am not sure if they are going the right direction, but it looks like that is their plan. The Itaniums are incredibly good at processing many different things at the same time and that is where Intel (and computers) seem to be going, heavily multitasked. All in all, it will make for an interesting few years for both companies.

Also, I see some people just think MS will release a 64 bit patch to Windows XP. I highly doubt they will just do that and not a whole OS for us to buy since that would be un-MS like. That will also hurt upgraders because if they go ahead and buy a AMD system, they will have to pluck down another $99-199 (depending on version) to get the 64 bit OS. I just see too many people putting faith in MS releasing a quick little patch that will magically update every single driver and files up to 64 bit ready when they are surely going to just release it as a separate version to make some more money. Okay, I am done, sorry for the ranting.

Jonny

 

[sonoran]

Originally posted by: jonny13
I just see too many people putting faith in MS releasing a quick little patch that will magically update every single driver and files up to 64 bit ready when they are surely going to just release it as a separate version to make some more money.

A patch to upgrade to x86-64? Fat chance! Like you say, they're gonna want to milk the cash cow.

 

[Jeff7181]

The greatest benefit to 64 bit computing is the amount of work done per clock cycle. You say we don't need to do more... I agree... but would it be nice if you had a 500 Mhz processor in your laptop, that could do everything your 2 Ghz processor could do, and do it using half the power? It's all about efficiency... maximizing the amount of work done per clock cycle. While the 500 vs 2000 Mhz example may be unrealistic, it's just an example to illistrate it for those of you who would say "there's not a significant different between 2 Ghz and 1.5 Ghz."
And while there may be architectural obstacles to overcome... more work per clock cycle means you don't have to increase clock speed as much to gain a lot of performance... that's what AMD has been doing for a while now with the Athlon XP line. So... If you can reduce the clock speed of a processor, yet keep the performance of it the same, or increase the performance, you extend the life of the processor. Lets say the current A64 architecture is physically limited to 2.6 Ghz. And lets say there's 2.6 Ghz A64's on the market, and AMD wants even better performing processors. They either A) figure out how to change the physical properties of the core to allow higher clock speeds... or B) make it do more work per clock cycle. With the Athlon-64, they've done both... SOI will allow for higher clock speeds, and the Athlon-64 can obvoiusly do more work per clock cycle... if you pair it up with an Athlon XP with the same clock speed it's easy to see that.

In summary... the benefits of AMD's 64-bit CPU:

1.) More work done per clock cycle, which allows lower clock speeds to get the same performance as current CPU's, which in my opinion, will extend the "life" of the processor architecture
2.) SOI reduces required voltage, which in turn should reduce heat
3.) Paves the way for future applications who can benefit from all the RAM 64-bit processor are capable of addressing


As far as x86-64 Windows XP... anybody who thinks a "patch" will be available does't know what they're talking about. It would be nearly impossible to patch existing version of Windows XP. Well, not impossible... but that would be highly inefficient since the entire kernel must be recompiled to use the x86-64 instructions... as well as changes made to the kernel to be optimized for those instructions. At least, that's my understanding... if someone else knows differently you're welcome to correct me.

 

[Mingon]

The greatest benefit to 64 bit computing is the amount of work done per clock cycle. You say we don't need to do more... I agree... but would it be nice if you had a 500 Mhz processor in your laptop, that could do everything your 2 Ghz processor could do, and do it using half the power?

But if more transistors are needed and more were used then the thermal characteristics would be equal for similar performance.

 

[Jeff7181]

Originally posted by: Mingon

The greatest benefit to 64 bit computing is the amount of work done per clock cycle. You say we don't need to do more... I agree... but would it be nice if you had a 500 Mhz processor in your laptop, that could do everything your 2 Ghz processor could do, and do it using half the power?

But if more transistors are needed and more were used then the thermal characteristics would be equal for similar performance.

True... but I don't think it's a direct correlation... if it was, P4's with HT would be creating 160+ watts... and that's just not happening.

 

[Mingon]

True... but I don't think it's a direct correlation... if it was, P4's with HT would be creating 160+ watts... and that's just not happening.

But the p4 only fills the remainding pipeline stages up, not a seperate pipeline. No doubt it would produce less heat - but on a similar manufacturing process its probably not much difference. look at the itanium, that is slower, with more transistors and that supposedly is a little bit toasty. I would have thought a Risc architecture would perhaps produce less heat tho.

 

[zephyrprime]

I would have thought a Risc architecture would perhaps produce less heat tho.

It doens't make a difference. Besides, both that athlon and pentium lines have been risc on the inside for a long time.

 

[zephyrprime]

The greatest benefit to 64 bit computing is the amount of work done per clock cycle.

I think you're thinking of superscaler features, not 64bittiness. The only time a 64bit processor does more work per cycle is when 64bit integer arithmatic is being done which isn't often. You cannot recompile/rewrite a program for 64bit and expect it to be anywhere near twice as fast.

 

[Mingon]

It doens't make a difference. Besides, both that athlon and pentium lines have been risc on the inside for a long time.

I thought they were CISC with some risc instructions e.g. sse

 

[Goi]

CISC and RISC has been quite irrelevant for quite some time. While the x86 ISA is considered CISC, it has employed many RISC like feature, such as:
- decoding the x86 instructions to RISCy micro-ops for the actual execution path to work on,
- superscaler execution
- pipelining
- branch prediction
- OOE
All of which are RISC features. I'm sure there are others, but off the top of my head those are what I remember.

 

[beyoku]

well i think they will be comparible. both will have a meg of cache but i dont know. I think AMD will come out on top kind of like ATI has been doing in the past year. What we may need to see is 64bit windoze and apps. WE do not yet know how those apps will run on the A64 - We dont even know how the 32bit apps will run under Win64 using WOW. They could blow prescott out the water. They could suck we dont know yet. Im going to love the battle though. I think im in AMD's corner though. TO me intel could be making products a lot better for the money that they have. It seems as if they would blow AMD away but no?

 

[Jeff7181]

TO me intel could be making products a lot better for the money that they have. It seems as if they would blow AMD away but no?

Competition is healthy, both Intel and AMD know that.