Naustica -
Why do you think this?
clarkey -
You need to note a few OTHER things...
1. The 90nm FX line isn't due until 2005 (it has been that way for some time)
2. The 4000+ wasn't due until sometime in Q4
3. It appears they are doing the same thing they did with Newcastle, i.e they are burning off the last of their 130nm inventory as locked and rebadged processors while they ramp up on the 90nm.
Why do you think this?
clarkey -
You need to note a few OTHER things...
1. The 90nm FX line isn't due until 2005 (it has been that way for some time)
2. The 4000+ wasn't due until sometime in Q4
3. It appears they are doing the same thing they did with Newcastle, i.e they are burning off the last of their 130nm inventory as locked and rebadged processors while they ramp up on the 90nm.
And to add to the confusion there is also the Opteron 150, which is also a S940 2.4GHz chip w/ 1MB of L2 (same as the FX-53), but has a locked multiplier and sells for around $640 or so at Newegg - about $120 cheaper than the S940 FX-53 and over $200 cheaper than the S939 version.
There are just too darn many 2.4GHz parts out there right now.
DrMrLordX
Lifer
- Apr 27, 2000
- 22,724
- 12,699
- 136
The Opteron 150 is lookin pretty good in light of all this. I must say, however, that it is somewhat disappointing thatAMD isn't pushing the performance envelope with their non-enthusiast parts. The FX-55 should keep them on top performance-wise, once it's out, but eh, 2005? Aren't they being a little sluggish? Who cares if Intel is having problems or not? All the more reason for them to put more distance between themselves and Intel.
I was told to believe the first 939 A64s were going on 90nm. But it's fine, I won't need to upgrade my mobo anytime soon since the new ones coming out including dual/multi-core will be compatible with all current 939 mobos.
Naustica - You really have to stop reading Ed's gibberish...he is operating under the false assumption that it's in AMD's best interest to turn out the fastest processors they can, as soon as they can. That would be disasterous for the company's bottom line! The best thing they can do for the company is to ONLY stay one step ahead of Intel, because that allows them to maximise their return...
As to yields from AMD being better than Intel or IBM, I will quote from the recent Goldman Sachs investigation...
As to yields from AMD being better than Intel or IBM, I will quote from the recent Goldman Sachs investigation...
Jeff7181
Lifer
- Aug 21, 2002
- 18,368
- 11
- 81
Not really too many 2.4 GHz parts... but definately too much price gouging. No way should a server level CPU cost less than a regular desktop processor. That would be like 3.0 GHz Xeon costing less than a 3.0 GHz Pentium 4... it's just crazy.
I just don't like the way things look... a socket 940 processor is cheaper than a socket 939 processor. It should be the other way around... socket 940 being more of a server type of platform since only the Opteron will continue to be made in the socket 940 variety. So people are staying away from socket 940 and going with socket 939 because they don't want to be locked into an Opteron only upgrade path. Even on the same socket, the server level CPU (Opteron 150) is less expensive than the desktop level CPU (FX-53). For all intents and purposes, they're identical processors... the only difference should be that the Opteron 150 has to go through a lot more testing and validation since servers demand 100% reliability.
It's just stupid... and I'm tired... so I'm going to watch Girls Gone Wild and go to bed.
When I built my computer I looked at my AMD64 processor and it actually says that it was manufactured in 2001 on it. So, if you think about it, AMD may already be making extremely fast procs that we don't even know about, and is just waiting to release them in accordance with Intels latest offerings in order to increase profits.
- Aug 22, 2001
- 31,743
- 31,694
- 146
This apears to be unconfirmed by AMD as of yet. Should it prove accurate then the PR isn't likely to hold up in many situations in a 32bit enviroment, but I read somewhere that in a 64bit enviroment the larger cache could be more useful. Perhaps it's with that in mind that they chose to rate it as such? Regardless, this wouldn't be the first AMD CPU to suffer from being the equivalent of an imaginary boost in horsepower by adding a racing sticker that sounds fast 
Prating is for common people. I don't care about the name of the processor or the Pratings, Before buying a processor I always evaluate the benchmarks scores and that's it. Besides both intel and AMD have confusing or wrong Ratings for their processors, common people don't know the difference btw P4A, P4B P4E, P4EE, and we all know for example that P4B 2.6 will be slower than P4C 2.4, but common people is not aware, so the will buy P4B 2.6 because they thing it is faster, when it isn't.
ponyo
Lifer
- Feb 14, 2002
- 19,688
- 2,811
- 126
I've read Ed for about 5 years now and I'll continue to read him. Why? Because he looks beyond PR bull most sites just repeats. While I may not always agree with him, I respect his views and opinions. Same reason I read Robert Cringely.
As for Goldman while I respect as a firm, I place about about much weight on any analcyst opinion as a big pile of horse manure. I've been trading actively for the past 11 years and been making a living as swing trader for the past 6 years and one thing I've learned is to fade most analcyst opinions if I want to make a profit.
Even with Intel lagging behind it is no reason to lay back and stretch the PR system, I believe and have good reasons to believe that they can get 2.6Ghz on 90 nm on a good yield, predictions are that with time 90 nm will allow speeds of 3Ghz or slight over.
512K more cache doesn?t equal 200+ points, I hope I am wrong but to me, with having seen benchmarks in the past, I don?t feel AMD can pull it off, sure 90 nm parts have these additional features :
90 nm improvments include:
Full SSE3 implementation
* Improved hardware data prefetch mechanism
* Increased number of writing combine buffers (D0 stepping A64's can now combine up to four non-cacheable streams compared to 2 on the C0 and CG stepping A64's)
* Improved on-die memory controller with more advanced open page policy
* On-die thermal throttling
* Black Diamond Low-K technology (slower less power hungry transistors in less used sections and faster and more power hungry transistors in frequently used sections of the cpu)
However I only believe these are minor improvements, but time will tell. Even if the 4000+ only had a 100 Mhz clock speed boost, It would be better then nothing and we all know how well an K8 responds to clockspeed boosts, more so then any netburst CPU ever has.
I think AMD are just too sporadic, I thought the point of going to 939 was to reduce cost of producing the clawhammer?s 1 Meg Cache and allowing the Newcastle dual channel memory controller into play, but they just throw these things out the door and decide to get the 1 Meg cahce back because they don?t want to use any of the clockspeed headroom which isn?t the same performance boost anyway
Ed over @ overclockers.com has written an article saying they cant make 2.6Ghz hammers on a 90 nm, and thats why, but I'll leave that up to you to decide if thats true.
512K more cache doesn?t equal 200+ points, I hope I am wrong but to me, with having seen benchmarks in the past, I don?t feel AMD can pull it off, sure 90 nm parts have these additional features :
90 nm improvments include:
Full SSE3 implementation
* Improved hardware data prefetch mechanism
* Increased number of writing combine buffers (D0 stepping A64's can now combine up to four non-cacheable streams compared to 2 on the C0 and CG stepping A64's)
* Improved on-die memory controller with more advanced open page policy
* On-die thermal throttling
* Black Diamond Low-K technology (slower less power hungry transistors in less used sections and faster and more power hungry transistors in frequently used sections of the cpu)
However I only believe these are minor improvements, but time will tell. Even if the 4000+ only had a 100 Mhz clock speed boost, It would be better then nothing and we all know how well an K8 responds to clockspeed boosts, more so then any netburst CPU ever has.
I think AMD are just too sporadic, I thought the point of going to 939 was to reduce cost of producing the clawhammer?s 1 Meg Cache and allowing the Newcastle dual channel memory controller into play, but they just throw these things out the door and decide to get the 1 Meg cahce back because they don?t want to use any of the clockspeed headroom which isn?t the same performance boost anyway
Ed over @ overclockers.com has written an article saying they cant make 2.6Ghz hammers on a 90 nm, and thats why, but I'll leave that up to you to decide if thats true.
DrMrLordX
Lifer
- Apr 27, 2000
- 22,724
- 12,699
- 136
Is it really? A single-cpu socket 940 system would only have two components that would cost more than in 939 system, and those would be the memory and probably motherboard. Namely, it'd have to be registered ECC RAM and a socket 940 board to match.
Would the price differential between Socket 939 board + non-registered RAM and socket 940 board + Registered RAM really exceed the price differential between the socket 939 FX-53 and the Opteron 150?
From Newegg:
S939 board price ~$120
S940 board price ~$175
Price Difference: ~$55
OCZ Enhanced Latency Dual Channel 512MB Kit: $117
OCZ Enhanced Latency Dual Channel ECC 512MB Kit: $144
Price Difference: $27
(Even if you go with 1GB kits, the difference only comes to $47 between the ECC and non-ECC Enhanced Latency kits).
So the difference in total cost between a S939 and S940 comes out to around $90 or so, depending upon the make and model of the boards you use for comparison. If you also take into consideration the fact that the Opteron platform will probably be a bit more stable if you seriously stress out the system on very RAM-intensive tasks thanks to the ECC, plus the fact that you have almost a guaranteed upgrade path to a dual core Opteron in the next year or so, you find that find a S940 system is actually a pretty good place to put your money right now if you are thowing together a high end system. The nForce 4 Pro S940 boards with SLI ought to make S940 an even more attractive option when they hit the streets - if the S939 processor prices don't drop down to below the Opteron 150 prices by then, of course.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter