Intel's Pentium Extreme Edition 955: 65nm, 4 threads and 376M transistors
by Anand Lal Shimpi on December 30, 2005 11:36 AM EST- Posted in
- CPUs
Larger L2, but no increase in latency?
When Prescott first got a 2MB L2 cache, we noticed that along with a larger L2 came a 17% increase in access latency. The end result was a mixed bag of performance, with some applications benefitting from the larger cache while others were hampered by the increase in L2 latency. Overall, the end result was that the two performance elements balanced each other out and Prescott 2M generally offered no real performance improvement over the 1MB version.
With Presler, each core also gets an upgraded 2MB cache, as compared to the 1MB L2 cache found in Smithfield. The upgrade is similar to what we saw with Prescott, so we assumed that along with a larger L2 cache per core, Presler's L2 cache also received an increase in L2 cache latency over Smithfield.
In order to confirm, we ran ScienceMark 2.0 and Cachemem:
What we found was extremely interesting; however, Presler does have the same 27 cycle L2 cache as Prescott 2M, but so does Smithfield. We simply took for granted that Smithfield was nothing more than two Prescott 1M cores put together, but this data shows us that Smithfield actually had the same higher latency L2 cache as Prescott 2M.
Although we were expecting Presler to give us a higher latency L2 over Smithfield, it looks like Smithfield actually had a higher latency L2 to begin with. This means that, at the same clock speed, Presler will be at least as fast as Smithfield, if not faster. Normally, we take for granted that a new core means better performance, but Intel has let us down in the past; luckily, this time we're not put in such a situation.
When Prescott first got a 2MB L2 cache, we noticed that along with a larger L2 came a 17% increase in access latency. The end result was a mixed bag of performance, with some applications benefitting from the larger cache while others were hampered by the increase in L2 latency. Overall, the end result was that the two performance elements balanced each other out and Prescott 2M generally offered no real performance improvement over the 1MB version.
With Presler, each core also gets an upgraded 2MB cache, as compared to the 1MB L2 cache found in Smithfield. The upgrade is similar to what we saw with Prescott, so we assumed that along with a larger L2 cache per core, Presler's L2 cache also received an increase in L2 cache latency over Smithfield.
In order to confirm, we ran ScienceMark 2.0 and Cachemem:
| Cachemem L2 Latency (128KB block, 64-byte stride) | ScienceMark L2 Latency (64-byte stride) | |
| AMD Athlon 64 X2 4800+ | 17 cycles | 17 cycles |
| Intel Smithfield 2.8GHz | 27 cycles | 27 cycles |
| Intel Presler 2.8GHz | 27 cycles | 27 cycles |
| Intel Prescott 2M | 27 cycles | 27 cycles |
| Intel Prescott 1M | 23 cycles | 23 cycles |
What we found was extremely interesting; however, Presler does have the same 27 cycle L2 cache as Prescott 2M, but so does Smithfield. We simply took for granted that Smithfield was nothing more than two Prescott 1M cores put together, but this data shows us that Smithfield actually had the same higher latency L2 cache as Prescott 2M.
Although we were expecting Presler to give us a higher latency L2 over Smithfield, it looks like Smithfield actually had a higher latency L2 to begin with. This means that, at the same clock speed, Presler will be at least as fast as Smithfield, if not faster. Normally, we take for granted that a new core means better performance, but Intel has let us down in the past; luckily, this time we're not put in such a situation.
Facebook
Twitter
RSS
84 Comments
View All Comments
skunkbuster - Friday, December 30, 2005 - link
cramitpal is that you?coldpower27 - Friday, December 30, 2005 - link
This is the Pentium Extreme Edition of course it's price is going to be 999US.If you want cheaper Presler cores, wait for the Pentium D 920 to 950 line to com out in Mid January.
Betwon - Friday, December 30, 2005 - link
INQ says Presler 920 will be about 240$.It is very interesting that PD820 defeat FX-57 in a SMP game.
phaxmohdem - Friday, December 30, 2005 - link
Let's recap, the X2 4800+ was ahead in most tests, and at worst could probably be called the 955 EE's equivilent....
955EE = $999
4800+ = ~$785
Yeah, I'd definately recommend "sticking with AMD for now and re-evaluating Intel's offerings when Conroe arrives."
Did anyone else notice how the lowly 3800+ did better in most gaming scenarios?
955EE = $999
3800+ = ~$315
Tasty :)
GhandiInstinct - Friday, December 30, 2005 - link
LOL, honestly your post is the only necessary post here. It compares and contrats the two perfectly in terms of which is a better buy, given the reader has seen all of the benchmarks in which the 4800 beats the 955EE.Intel just can't win because of EGO.
Anemone - Friday, December 30, 2005 - link
Kudos because no matter where you sat personally you seemed to have called the shots fairly. I'd agree with the conclusion as well, that you are either Conroe or A64, that the P4 is an overdue dead end. It performs well, but it is hot and uses lots of electricity to do so. Overclocking wasn't needed because, quite frankly the X2 chips oc too, and you'd find they probably do it better.Socket M2 is again, something you "should" wait for if you can, as is Conroe. These are heavy recomendations, you really would be very smart to wait for these two things. Barring that, given the better of two bad options (meaning you have to upgrade now when you should be waiting), AMD is the better choice, partially for the power consumption, partially for the "less of a dead end than a P4" issue.
Still, heavy, heavy emphasis on "you should wait", as a complete changeover is going on with both AMD and Intel and your ability to perform minor upgrades 1-2 years from now will depend on waiting patiently for a few more months.
:)
JarredWalton - Friday, December 30, 2005 - link
Socket M2 doesn't appear to be anything special. Why wait 6 months for a 5% performance boost and a RAM change? Just like waiting for Prescott ended up being much ado about nothing, M2 isn't going to be wildly different from today's 939 chips. Get a good socket 939 system with an X2 and SLI, and you should be set for at least 18 months.Calin - Tuesday, January 3, 2006 - link
I don't find SLI important - except for the possibility to run two top of the line video cards. And increased speed won't come from higher RAM speed - not so much anyway in order to keep you waiting.I just wonder how long will the Socket 939 be kept - considering that the value line is the cheaper Socket 754 (cheaper in having a single memory channel, so half as many lines to memory banks). Or if Socket 754 will be abandoned before Socket 939, or if Socket M2/2 (single channel DDR2 memory) will appear.
nserra - Friday, December 30, 2005 - link
Two cores on the same packing is an excellent idea!Will amd do that with m2?
Could lower the dual core price and even at 90nm could put 2 dual core processors on the same packing and build a 4 core processor (fake one, but 4 cores there).
ViRGE - Friday, December 30, 2005 - link
The problem with 2 physical cores is that you're forgoing any sort of on-die communication benefits by doing so. It's certainly cheaper for Intel to make things this way, but it's a poor way to go for performance, as it makes it harder for the cores to quickly send data to each other and share resources. It's certainly a valid solution(especially given how Intel didn't have any inter-core communication even when both cores were on the same die), but ultimately a combined die for inter-core communication is superior for performance and scaling.