Sitting in a seemingly endless number of meetings at CES, you quickly realize that keeping up with all of the product releases, socket flavors, chipsets to wait for and GPUs to lust after is virtually a full time job.

We try and ease the burden as much as possible by providing you with coverage as soon as we get the information, but what we're seeing at the start of this new year is that with so much changing it's necessary to work a little harder to simplify things. The Intel world is pretty easy to follow; Prescott is due out soon in a Socket-478 flavor, followed by a LGA-775 version. By the end of this year Intel may begin sampling Tejas, the successor to Prescott but very little is known about the chip. Tejas has been rumored to be a multicore desktop chip, however we're now receiving information contrary to what we had once thought.

On the chipset side, we are all waiting for Grantsdale and Alderwood from Intel, the successors to the 865 and 875 platforms. But where Intel's 2004 roadmap is pretty straight forward, AMD's is a bit more cluttered with the confusion of varying cache sizes, Socket-939 and model numbers. Although most of the information about AMD's future plans have already been leaked on the net, we decided to put together a quick reference page to AMD's 2004 CPU roadmap to go along with our recent coverage of chipset plans.

The roadmaps are divided according to CPU socket and the rest is self explanatory - we hope the information is concise and useful:

Socket-754 Roadmap for 2004
  Clock Speed Cache Size Release Date
AMD Athlon 64 3700+
2.4GHz
1MB
Q2 '04
AMD Athlon 64 3400+
2.4GHz
512KB
Q2 '04
AMD Athlon 64 3400+
2.2GHz
1MB
Already Available
AMD Athlon 64 3200+
2.0GHz
1MB
Already Available
AMD Athlon 64 3000+
2.0GHz
512KB
Already Available
AMD Athlon XP 3000+
TBD
256KB
Q4 '04
AMD Athlon XP 2800+
TBD
256KB
Q3 '04

As of now, the fastest Socket-754 CPU will be the Athlon 64 3700+; current motherboard owners looking for an upgrade path can look no further than the 3700+ running at 2.4GHz.

What's interesting is that the model numbers seem to scale more than linearly with clock speed, something we noticed with the Athlon XP that eventually led to its misleading model numbers. In the case of the 3400+, a 10% increase in clock speed over the 3200+ resulted in a 6.25% increase in the model number (resulting in the 3400+). However, if we look at the 3700+, a 9.1% gain in clock speed results in an 8.8% increase in model number. Where we cannot draw a parallel to the Athlon XP model number situation is in the fact that the Athlon 64's on-die memory controller does allow the CPU to scale much better with clock speed. It could very well be that AMD is counting on the Athlon 64's performance scaling much better with clock speed than Prescott, thus justifying the increased model numbers.

What's also worth mentioning is that the 3700+ will be the last 1MB L2 cache Athlon 64 to hit the market, every other Athlon 64 will feature a 512KB L2 cache. The smaller cache size brings us to the difference between the two Athlon 64 3400+ processors listed in the chart above; the current 2.2GHz 3400+, as you know, features a 1MB L2 cache. In the next quarter, AMD will bump the clock speed of the 3400+ to 2.4GHz and cut the cache in half in order to maintain the performance rating.

The last thing to note are the two Socket-754 Athlon XPs on the chart; although clock speeds have yet to be determined, you can expect these two processors to feature an on-die memory controller just like their Athlon 64 brothers but have their 64-bit support disabled. The reduction in cache size to 256KB will decrease 32-bit performance by a noticeable amount, but they will make a good successor to the current Socket-A Athlon XPs.

Socket-939 Roadmap for 2004
  Clock Speed Cache Size Release Date
AMD Athlon 64 FX-55
2.6GHz
1MB
Q4 '04
AMD Athlon 64 FX-53
2.4GHz
1MB
Q2 '04
AMD Athlon 64 4000+
2.6GHz
512KB
Q4 '04
AMD Athlon 64 3700+
2.4GHz
512KB
Q4 '04
AMD Athlon 64 3700+
2.4GHz
512KB
Q2 '04
AMD Athlon 64 3400+
2.2GHz
512KB
Q2 '04

The new Socket-939 platform will bring dual channel support to the entire line of 64-bit AMD processors, and it will also take away half the cache from all but the FX processors. Although the Socket-754 3700+ features a full 1MB L2 cache, the Socket-939 version will not be given more than 512KB. What's interesting is that the two processors carry the same 3700+ model number, running at the same 2.4GHz, while the Socket-754 version features twice the cache. AMD seems to be indicating that the advantage of a 128-bit memory interface will offset any performance loss incurred by halving the cache. We have not seen much data to support this theory, but we'll keep a close watch on it.

If you'll notice, there are two 3700+ processors on the roadmap above and since we're trying to clear up confusion, here's an explanation: the Q4 Socket-939 processors are supposed to be based on AMD's 90nm process (that includes the 4000+ and the FX-55). AMD has demoed 90nm Opterons already, but they were running at 800MHz so there's no indication of how well AMD will be able to stick to this roadmap. If we were to expect any deviation from the roadmap as it stands it would be in Q4; new manufacturing processes are not easy to ramp up as we've seen time and time again from both AMD and Intel (Prescott anyone?).

Socket-940 Roadmap for 2004
  Clock Speed Cache Size Release Date
AMD Athlon 64 FX-53
2.4GHz
1MB
Q1 '04
AMD Athlon 64 FX-51
2.2GHz
1MB
Already Available

Finally we have the quickly demising (at least on the desktop) Socket-940 platform, the FX-53 looks like the end of the line.

We hope this has been helpful; if you'd like to see similar quick reference articles about other companies let us know.

POST A COMMENT

60 Comments

View All Comments

  • MoronBasher - Thursday, January 15, 2004 - link

    Rustang you are wrong.

    Really, you are. The FX is a rebadged opteron. For those who has an FX right now, they can always put their cpus on Opteron mobos in the future. and upgrade through that. So amd really isn't really leaving anyone in the dust as many have said. All that amd is going to take away is the name off of the "athlon 64 fx" and put back the proper name of opteron in its line up

    Think b4 you speak (in this case... type)
    Reply
  • Rustang - Thursday, January 15, 2004 - link

    So what happens to all the AMD enthusiasts and fans who shelled out the $,$$$ to buy a top of the line FX-51 platform? No 1GHz 250 chipset and no future path for upgrades? That is more than irresponsible... AMD should allow owners of 940 Athlons to to the 939 chip FOR FREE.

    If that doesn't happen then AMD has lost my (and I'm sure a whole lot of other people's) business. Next time I'll spend my money on an overpriced Intel knowing that 1-2 years from now I can still upgrade the CPU/mobo and be up to date, rather than get an AMD solution. There's something seriously wrong with the picture when the "flagship" FX-51 platform gets retired by AMD within 6 months (give or take) of introduction and when you need to be a soothsayer to know that the company will stiff you and not support the product. The moment you become customer complacent, is the moment you become fat and overrated/priced like Intel...
    Reply
  • Ilmater - Thursday, January 15, 2004 - link

    sajidk1975: That changes all the time, and roadmaps are hard enough to get information on, MUCH LESS the actual pricing. Plus, without the processors in hand, how can anyone possibly know the best price/performance now?

    Insomniac: Yeah, I would like to see those as well. I just can't keep up with this stuff, and I'm CONSTANTLY in here.

    PrinceGaz: Good Lord yes. This was a GREAT idea and a wonderful quick reference, but that balck text on dark green part gives me a headache to look at. PLEASE CHANGE THAT SOMEBODY!
    Reply
  • TrogdorJW - Wednesday, January 14, 2004 - link

    Oops... screwed up on my math a little bit in the above. For AMD, I should have been calculating 16 million transistors per 256K of L2, and at some point I started doing it as 17 million. Figures might be off by a few million, then, but overall the point still stands. (And since I was estimating anyway, who cares? Well, besides me.) Reply
  • TrogdorJW - Wednesday, January 14, 2004 - link

    According to the following CPU comparison:

    http://www.tomshardware.com/cpu/20031223/images/im...

    If you look at the Athlon XP, there are 54.3 million transistors in the Barton. There are 37.5 million in the Thoroughbred. So doubling the L2 cache from 256K to 512K required an extra 16 million transistors. On the Intel side, they list a P4's L2 cache as taking up 26 million transistors for 512K. Given the exclusive nature of the L2 cache on Athlon CPUs, I wouldn't be surprised to seem them use more transistors relative to the size of the cache. So the Barton might be 34 million transistors for the cache and 20 million for the CPU logic. (The original Athlon was 22 million for the CPU, with the cache located elsewhere, so this is pretty accurate.)

    Now, if you look at the Athlon 64/FX/Opteron, you will see that it is currently taking 105.9 million transistors. That's with 128K L1 and 1024K L2. If the numbers are similar to the Athlon XP (and they most likely are), the L2 cache is using up roung 68 million of the transistors. That leaves 38 million for the CPU core. With that additional 64-bit circuitry, registers, SSE2, and memory controller, that seems pretty likely. Take the 22 million transistors of the Athlon, and use another 16 million adding those features.

    So, if you agree with all of that so far, then create an Athlon XP CPU based off the K8 core. So start at 106 million transistors. Cutting out 768K of the 1024K L2 should eliminate roughly 51 million transistors, and we're down to 55 million (which is the size of current Barton CPUs). Knock off the 64-bit support, and you might kill off another 8 million transistors, give or take. All told, you end up with a CPU that uses maybe 45 to 50 million transistors.

    If that's correct, the new Athlon XP for socket 754 should cost LESS to produce than the current Barton. AMD can also stop production on the Barton and forget about socket A. The yield of the "Paris" CPU (I believe that's the codename) would be much higher than the Athlon 64/FX/Opteron CPUs. They also wouldn't need to worry about shifting it to the 90 nm process, so that would save them some money.

    Final tally, then, is that they get a CPU which should roughly equal the current Athlon XP 3200+. (I'm figuring that it should be clocked at least 2.0 GHz, and while it has half the L2 cache of the Barton, the integrated memory controller should roughly equal out.)

    It's not really a bad business move. Revamping the current locations for 90 nm is pretty much out of the question - they were built for 130 nm with 200 mm wafers, and replacing the equipment so that they could do 90 nm would cost almost as much as building a new plant. It should be a good budget CPU, certainly better than the Celeron CPUs that Intel is currently peddling.

    That's my take. Still hate the PR names, though.
    Reply
  • KF - Wednesday, January 14, 2004 - link

    If AMD sticks to this timetable, AMD has planned on getting killed in top-of-the-line 32 bit performance over the next year. The only thing good about it is they probably won't fall short. (Falling short is really bad for PR and stock price.) Intel's HT performance with SSE2 is probably going boost dramatically with Intel's Prescott, because that is probably what all the addition die circuitry is for, which will make A64 look very bad in encoding benchmarks, which A64 already loses. The only question is how fast Intel is going to drop the price premium of the Prescott. Since the economy is in an upswing, and Intel is getting pretty tired of selling good chips under $200, don't expect much mercy. In the unlikely event MS gets A64 Windows out the door, that may turn things around for AMD.

    On the surface of it, if they plan to cripple A64s into A32s, AMD should fire their whole marketing department. Unless this is going to reduce PRODUCTION costs to the point they can make money selling SOI chips at say $120, what is the point? Don't give me that stuff about salvaging defective chips. If 95% of the chips you sell are saved rejects, you are wasting money by putting the extra parts on the die. Maybe at $120 calling the chips XPs and having them plug into cheap A64 mobos could produce a reasonable "value" combo between real XPs and A64s, but there isn't a lot of room. Or do they plan on making SOI chips more cheaply than the conventional ones?

    It also looks like the A64 performance boost is not much to wait for if you plan on getting an A64 (other than the socket switch.) A64 3000+ is not that far short of what is going to come along, and it seems doubtful the prices will drop so much that it wouldn't be better to buy now and get the benefits sooner.

    That's if you believe the timetable. Maybe AMD has just switched tactics from one extreme to the other. No info about future performance. When it arrives, you will see it for sale first. Touting the future has never helped AMD. Sandbagging Intel worked great for the Athlon's intro. If that thing about crippling A64s to A32 is true, then AMD must think they can get the SOI chip price down near what current chips cost. Otherwise why?
    Reply
  • Shinei - Wednesday, January 14, 2004 - link

    They wouldn't; hence, marketing was born. Reply
  • jeromekwok - Wednesday, January 14, 2004 - link

    The relationship between model number and clock frequency/caches is very confusing.

    AMD want all of us forget clock freq and caches, and stick with model number. So is it good?

    If VIA release a Cyrix 5000+ 1GHz 128kB CPU, how could a consumer know Athlon 64 4000+ is faster than Cyrix 5000+?
    Reply
  • Ajay - Wednesday, January 14, 2004 - link

    Hmm, I don't get why AMD would be truncating s754 cpus so early - they are a great upgrade to the XP line. Seems like they are just trying to keep processor prices up.

    I also don't get why they are introducing _low_ _end_ XPs onto s754. "The reduction in cache size to 256KB will decrease 32-bit performance by a noticeable amount, but they will make a good successor to the current Socket-A Athlon XPs."

    So they'll be slower than the current XP 3200+...what kind of successor is that!!!

    Reply
  • PrinceGaz - Wednesday, January 14, 2004 - link

    @marky1124: Tom's recently did a round-up of all the gfx-cards from the past couple of years so you can compare the performance. Obviouslly it doesn't have R4xx or NV4x yet of course but its a good starting-point to compare future results with. Just remember to ignore the bit at the end where they add up all the fps scores to come to a conclusion as thats almost meaningless as it gives Q3 and other old/fast games too much importance; why they didn't normalise the results based on a particular card first is beyond me. Reply

Log in

Don't have an account? Sign up now