• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

First S939 Review!!

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Gautam

Senior Benchmark Addict
Joined
Feb 4, 2003
Location
SF Bay Area
Here

And, it looks just as lackluster as I thought. In gaming, the 3500+ actually does worse than the 3400+ on virtue of the cache. Puts out a strong suit in DiVX encoding, but this socket really isn't anything revolutionary. For the price, it looks like it would be a worthless premium for most.
 
I dunno, maybe it's because I've been a French student for 5 years now, but I don't think there's anything too confusing about it. ;)

What's noteworthy:

Startling 93% memory bandwidth efficiency, 20% ahead of P4's. AMD has the lead in the raw bandwidth area quite handily now. The memory controller truly is a work of art. They've scarcely lost 2-3% compared to single channel.

In SuperPI, the 2.2ghz/512k 939 is equal to a 2.2ghz/1mb 754.

In all the actual games, the Clawhammer is in fact superior to the Newcastle, just as I had forecasted.

Seeing this, I'm beginning to think I may not bother to upgrade to 939. I was considering it for a while, but this doesn't look promising enough.
 
Let's take a look at this comparison...

The 754 Pin 3400+ was reviewed on a very mature Asus K8V with a total of 7 bios versions under its rom and mature K8T800 drivers from VIA.

The 939 pin 3500+ was reviewed on a pre-release Asus A8V which does not even apear on the Asus web site yet and in all likelyhood has a beta bios (speculation) and since it uses the new K8T800 Pro chipset, the drivers still need tweaking no doubt.

The 3400+ wins the Antique gaming benchmarks (Quake 3 will run entirely from the L3 cache on a P4EE) and other gaming benchmarks are esentially a dead heat.
 
This is about the 939 memory bandwidth I put up a month ago, before the 939 test results. It was based on analysing A64 FX- and 754-data. I think it turns out to be accurate.

Originally posted by hitechjb1

This 939 platform memory bandwidth, as estimated from some test data (so result is preliminary), is impressive. Its efficiency is around 86-90%, which is 15-20% (to be confirmed with more 939 test data) better than the P4 QDR dual channel counterpart.

Its effective bandwidth (not max), running at the same memory bus speed, is about 15-20% higher than that of P4 QDR dual channel and 81-89% higher than that of 754 platform or nforce2 dual channel.

Estimation and importance of 939 platform memory bandwidth

A major difference between the AMD 754 and 939 platforms is the memory bus, i.e. 64-bit memory bus for 754 vs the 128-bit memory bus for 939. Here put it some estimate (since 939 is not commonly available yet) to see the potential impact on memory bandwidth performance.

I think there is a significant advantage from the 939 128-bit memory bus and on-chip dual channel controller, it is very different from the nforce2 dual channel which has only few % memory bandwidth improvement over single channel, as shown below.

memory_bandwidth_efficiency = effective_memory_bandwidth / max_memory_bandwidth

1. In the P4 arena, the dual channel QDR efficiency is around 75% with 64-bit memory bus
max_memory_bandwidth = FSB x 4 x 8 = 32 FSB
effective_memory_bandwidth = FSB x 4 x 8 MB/s x 0.75 ~ 24 FSB

2. XP nforce2 single channel efficiency ~ 85-90%
max_memory_bandwidth = FSB x 2 x 8 = 16 FSB
effective bandwidth = 0.875 x 2 x 8 x FSB ~ 14 FSB

3. XP nforce2 dual channel effieiency ~ 90 - 95% (actually should be 45-48%, depends on how it is counted)
max_memory_bandwidth = FSB x 2 x 8 x 2 = 32 FSB
max_FSB_bandwidth = FSB x 2 x 8 = 16 FSB (FSB limits dual channel memory bandwidth)
effective bandwidth = 0.925 x 2 x 8 x FSB ~ 14.8 FSB

4. 754 hardwares have been around for a while, and we have seen its memory bandwidth being around 95%.

For 754 platform, memory bandwidth efficiency ~ 95%
max_memory_bandwidth_754 = 2 x 8 x memory_bus_frequency = 16 memory_bus_frequency
effective bandwidth = 0.95 x 2 x 8 x memory_bus_frequency = 15.2 memory_bus_frequency

E.g. from Maxvla's system screenshot (http://www.maxvla.com/host/komusa4200b.jpg), a 754 memory benchmark (integer buffered iSSE2) shows the 754 memory efficiency being around 4574/4800 = 95%.
At 300 MHz, the max bandwidth would be 4800 MB/s for single channel, and 9600 MB/s for 128-bit bus (theoretical max).

5. For the 939 128-bit memory bus, there is a good possibility that it could be higher than 75% (the P4 QDR number) due to its direct 128-bit memory bus:
- max_memory_bandwidth_939 = 2 x 16 x memory_bus_frequency = 32 memory_bus_frequency
- At 80%, 300 MHz, the effective bandwidth would be 7680 MB/s
- At 90%, 300 MHz, the effective bandwdith would be 8640 MB/s
- At 95%, 300 MHz, the effective bandwdith would be 9120 MB/s
(ECC is not required in 939).

I think the 128-bit memory bus could be more efficient than the 64-bit QDR, hope it is close to the single channel number ~ 85-90%. This will be confirmed when actual 939 hardwares come out. (Will see)

...


Summary (preliminary numbers, may vary as more 939 test results become available):

- If further confirmed by more 939 hardwares, this 86 - 90% number on bandwidth efficiency for 939 128-bit is 15 - 20% higher than the 75% QDR of P4 (64-bit).

- At 86-90% efficiency, the effective bandwidth for the 939 128-bit memory bus would be 81 - 89% higher than that of a 754 64-bit memory bus, with assumed 95% memory efficiency.

This higher bandwidth in 939 would have significant impact on memory intensive applications such as video and image streaming, applications using spatially structured data as in scientific computation, ..., as well as 3Dmark01.


For complete analysis,

Estimation and importance of 939 platform memory bandwidth (page 19)


Related links:

Differences between the XP FSB and the A64 buses (separate memory bus and HyperTransport bus) (page 19)

Some remarks on cache latency, cache size, memory latecny and memory bandwidth (for A64's) (page 19)
 
Last edited:
Regarding to overall performance comparison between 754, 939 and 940 platform, ....

From
A64 CPUs, chipsets, motherboards

On performance difference

The numbers on performance made here are based on two benchmark results and analysis of some A64 FX, A64 754 ClawHammer (1MB L2), A64 754 NewCastle (512KB L2), Barton, P4's, ...

http://www.ocforums.com/showthread.php?s=&postid=2762781#post2762781
http://www.ocforums.com/showthread.php?s=&postid=2766934#post2766934

For 939 with 1 MB L2, when available, it would perform similar to a 940 FX/Opteron, running at the same frequenies.

For 939, if used with unbuffered memory module, can be slightly faster than a 940 due to less memory overhead.

Both of 939/940 dual channel have 80% more effective memory bandwidth than the 754, hence they are always better in performance, especially for memory intensive programs, up to 20-80% higher performance.

From a few gaming benchmarks, a A64 FX/939 at 2.4 GHz performs 12-20% better than an A64 754 with 1MB L2 at 2.0 GHz, and 15-29% better than an A64 754 with 512 L2 at 2.0 GHz (memory bus, HT bus same frequencies). Not clear if 754 CPU's were clocked to same speed, what would the performance difference be, as the performance difference can be attributed to both memory bandwidth and CPU raw power, but these numbers put an upper bound on gaming performance of 939 over 754. From looking at another set of game benchmarks with both a 939 (512 KB L2) and a 754 (1 MB L2), both running at same frequencies, they are about tie.

So I can conclude that for gaming, if both 939 and 754 have the same L2 cache size, running at the same frequencies of CPU, memory and HT, the 939 performs few % (say 5%) on the average better than a 754 for most games.

If not counting memory intensive programs, the advantage of 939/FX to 754 is only few % (say 5%) on the average.

Running at the same frequencies of CPU, memory (and bandwidth), HT, the performance difference between a 512 KB L2 and a 1 MB L2 would average around 5%, between 2-10%+ depending on the application.


So what system would be built?
 
Please note that in that article comparison, the two 939 have only 512 KB L2 whereas the two 754 have 1 MB L2.

It is known that twice the size of L2 have few % performance advantage on the average for many programs.

That is also why ClawHammer (1 MB L2) performs few % better than NewCastle (512 KB L2) on the average for system running same frequencies of CPU, memory bus and HT bus.
 
Guatam, no offense man but u remind me of that little annoying kid that everyone knew in elementary school that always thought the things he had were better than everyone elses and that nothing else was better, no matter what
 
Well, the 939 is still in beta stage, or prototype stage. Still too early to tell how good it is. We should wait a little longer before drawing any conclusions ourselves.
 
From a few gaming benchmarks, a A64 FX/939 at 2.4 GHz performs 12-20% better than an A64 754 with 1MB L2 at 2.0 GHz, and 15-29% better than an A64 754 with 512 L2 at 2.0 GHz (memory bus, HT bus same frequencies). Not clear if 754 CPU's were clocked to same speed, what would the performance difference be, as the performance difference can be attributed to both memory bandwidth and CPU raw power, but these numbers put an upper bound on gaming performance of 939 over 754. From looking at another set of game benchmarks with both a 939 (512 KB L2) and a 754 (1 MB L2), both running at same frequencies, they are about tie.
We'll see this with the 2.4GHz 3700+.

The 939 isn't in the beta stage. It's very well developed, not only that, but it will officially be in stores on June 1st. It's simply a version of the year-old socket 940 platform that doesn't require registered memory. There isn't anything negative about it. It performs just as well as the architecture would lend itself to doing in theory, even better perhaps.
 
hitechjb1 said:
Please note that in that article comparison, the two 939 have only 512 KB L2 whereas the two 754 have 1 MB L2.

It is known that twice the size of L2 have few % performance advantage on the average for many programs.

That is also why ClawHammer (1 MB L2) performs few % better than NewCastle (512 KB L2) on the average for system running same frequencies of CPU, memory bus and HT bus.
However, the question of the practicality is raised for the everyday consumer in regards to the $700+ 1mb L2 DC-enabled FX-53... Initially, the consumer may have the choice, realistically, between the 1mb SC Clawhammer, and 512k DC Newcastle. The Clawhammer option will probably disappear in a couple of weeks.
 
Do you mean we won't be able to buy A64 754 with 1 MB L2 in a couple of weeks, such as A64 754 3200+, ...? What do you mean?
 
Except for the mobile versions, yes. All of the latest 3200+'s appear to be NewCastles. The Clawhammer 3200+'s are no longer ubiquitous in the wild. There is also much speculation about a 512k 2.4GHz part for the new 3400+.
 
Gautam said:
Except for the mobile versions, yes. All of the latest 3200+'s appear to be NewCastles. The Clawhammer 3200+'s are no longer ubiquitous in the wild. There is also much speculation about a 512k 2.4GHz part for the new 3400+.

I thought AMD just released a few 1.4V Mobile A64 754 CPU model 2800+, 3000+, 3200+ with 1 MB L2, which should be based on ClawHammer core.

Also there are the 3000+, 3200+, 3400+ Mobile DTR models with 1 MB L2 (ClawHammer).

Are you implying they will be gone soon?
Are they official findings?
 
Actually socket939's 90 nm cores will support SSE3 among other things. It does bring new stuff, along with new power specifications :p
 
So it pretty much seems like s939 is just slightly better than s754 making marginal gains which should just about be expected normally. The only considerable advantage of going 939 is really upgradeability into the future. A 939 mobo has a much better chance at outlasting a s754 mobo.

silentfire said:
Guatam, no offense man but u remind me of that little annoying kid that everyone knew in elementary school that always thought the things he had were better than everyone elses and that nothing else was better, no matter what

No offense man, but this post reminds me of posts made by members who ended up being banned from this forum. Gautum contributes a lot of information and he's helped me and others find a lot out about s754 and s939... Regardless of that however, your post could be interpreted as a personal attack and that is not encouraged or allowed at this forum, in any form.

I encourage everyone to keep this in mind, and be aware that comments like these DO NOT go unnoticed - I appreciate everyone keeping this thread on topic.
 
hitechjb1 said:


I thought AMD just released a few 1.4V Mobile A64 754 CPU model 2800+, 3000+, 3200+ with 1 MB L2, which should be based on ClawHammer core.

Also there are the 3000+, 3200+, 3400+ Mobile DTR models with 1 MB L2 (ClawHammer).

Are you implying they will be gone soon?
Are they official findings?
The official finding is that the CG Clawhammer core is discontinued. OC Detective posted it in my results thread, I gotta get going though, sorry.
 
Come on guys...
HERE

Gautam you are just a bitter, bitter, Socket 754 owner ;) Anyways...it's not looking too good. In benchmarks, the Socket 754 A64 3400+ beats the similarly clocked Socket 939 A64 3500+ almost every time. Another promising note...in Sandra memory benchmark the A64 3800+ beats the FX-53 by 8%. Even with a Socket 939 FX-53 gaining 5-10% because of using unbuffered dimms, close FX-53 performance can be had with less money with the A64 3800+ (assuming that the loss of cache is less felt as AMD scales up). Still, priced at a premium, this just makes me want to stay put.

EDIT: Wait, Gautam so you think chips like this will be discontinued in favor of Socket 754/939 Newcastle cores?
 
Last edited:
Hey, I wanted to buy a 939. That's the plan I came up with in the shower this morning, come back a couple hours later and see this. ;) I'd be less bitter if everyone didn't tell me I was nuts for going A64 in the first place. :D

Nevermind the discontinuing of the CG's. That only referred to the 2700+ mobile, I think. There hasn't been anything official about that, I was wrong. But what we are seeing are lots more 754 Newcastles in the wild, and a lot less Clawhammers. And it would make more sense for mobile processors to pack less cache. If you notice the week numbers on the mobiles, they're all pretty early. All less than the 10th week of 04, AFAIK. There aren't any brand new, fresh Clawhammers that I've personally seen, but several new Newcastles. In the end, it all boils down to prices and yields.
 
At least the mobile a64 clawhammers will be around for a while. I'll probably sell my 3000+ and get a clawhammer in a while.
 
Back