What are the basic differences between Prescot and Northwood?
- Thread starter videopho
- Start date
Williamette. Original P4, 256k cache, 180nm. They were pretty slow, and ran a bit warm as well, not as bad as prescott. Generaly the higher end P3's peformed better than williamette did. Socket 423 and 478 variations.
Northwood: 130nm, 512kb cache, 400, 533, and 800mhz FSB variants. 800mhz FSB versions, and the 3.06ghz 533mhz FSB chip have hyperthreading. Socket 423 and 478 variations.
Prescott: 90nm, 1mb and 2mb cache versions. Mainly 800mhz FSB with hyperthreading. 5xx series has 1mb cache, no 64bit(there are some 5xxj's which have execute disable, and some rare 5x1's that have 64bit). 6xx has 2mb cache, and 64bit. There are a few 5xx series, with no Hyperthreading that run on 533mhz FSB. Prescott had enhancements to hyperthreading, and SSE3 added. Also increased pipeline, they run pretty hot, and clock for clock, are a bit slower than northwood due to the increased pipeline length, and while they have more cache, the cache is slower than northwoods. Socket 478 and LGA775 variations.
Cedar Mill: 65nm, die shrunk prescott, 6x1 series. 2mb cache, 64bit, some have vanderpool(virtualization) I believe. Run cooler than prescott by a decent amount, and have been known to overclock well. LGA775 only.
Northwood: 130nm, 512kb cache, 400, 533, and 800mhz FSB variants. 800mhz FSB versions, and the 3.06ghz 533mhz FSB chip have hyperthreading. Socket 423 and 478 variations.
Prescott: 90nm, 1mb and 2mb cache versions. Mainly 800mhz FSB with hyperthreading. 5xx series has 1mb cache, no 64bit(there are some 5xxj's which have execute disable, and some rare 5x1's that have 64bit). 6xx has 2mb cache, and 64bit. There are a few 5xx series, with no Hyperthreading that run on 533mhz FSB. Prescott had enhancements to hyperthreading, and SSE3 added. Also increased pipeline, they run pretty hot, and clock for clock, are a bit slower than northwood due to the increased pipeline length, and while they have more cache, the cache is slower than northwoods. Socket 478 and LGA775 variations.
Cedar Mill: 65nm, die shrunk prescott, 6x1 series. 2mb cache, 64bit, some have vanderpool(virtualization) I believe. Run cooler than prescott by a decent amount, and have been known to overclock well. LGA775 only.
With all due respect, the usual blah blah about heat, power and performance per clock is misleading without also including the fact that the Prescott naturally became cheaper and capable of higher clocks and thus performance and also notably did not suffer the well known Northwood Sudden Death Syndrome which I was unfortunate to experience.
Prescott was designed for higher clocks and even if it did not achieve half of what the brainiacs at Intel counted on, it was still an advance as much as most other architecture changes had been. In comparison, the first P4's were kinda crap compared to the last PIII's but that disparity did not last long (unless perhaps clinging to the unrealistic clock for clock, cache for cache argument).
Niggly performance and architecture aside, it is likewise akin to comparing Coppermine to Tualatin in that respect. At some point, the previous generation hits a performance and EOL wall yet prices do not drop since they will remain the only upgrade option for many boards. If the possibility exists to upgrade to the new generation CPU in the same system then it is by far the best option.
When my Northwood 2.4C died a replacement would have cost slightly more than a Prescott 3.0 while the Northwood 3.0 was substantially higher and could not achieve the same clocks (and certainly not without risking NSDS). Indeed, before finding a refined P4 model "E" Prescott, I found the Celeron D (Prescott) 2.4 preferable to the Northwood for the short term since it was a fraction of the price and performed virtually the same when not multi-tasking (since it lacked HT).
Here are some old 3DM01 benchmarks with a Radeon "9700" (9500 256-bit 8pp). The first is at default viddy clocks while the rest are o'erclocked. The drivers vary but it still roughly demonstrates the differences. The main point being that the comparable Northwood simply could not compete with the Prescott on price and even if it could was not guaranteed to so effortlessly achieve the same performance (at default voltage no less). So, while technical arguments are fun they don't necessarily apply in practice.
P4 2.4
10561
P4 2.4 @ 2900
15834
C 2.4 @ 3060
15945
P4 3.0 @ 3014
16896
P4 3.0 @ 3608
18509
Prescott was designed for higher clocks and even if it did not achieve half of what the brainiacs at Intel counted on, it was still an advance as much as most other architecture changes had been. In comparison, the first P4's were kinda crap compared to the last PIII's but that disparity did not last long (unless perhaps clinging to the unrealistic clock for clock, cache for cache argument).
Niggly performance and architecture aside, it is likewise akin to comparing Coppermine to Tualatin in that respect. At some point, the previous generation hits a performance and EOL wall yet prices do not drop since they will remain the only upgrade option for many boards. If the possibility exists to upgrade to the new generation CPU in the same system then it is by far the best option.
When my Northwood 2.4C died a replacement would have cost slightly more than a Prescott 3.0 while the Northwood 3.0 was substantially higher and could not achieve the same clocks (and certainly not without risking NSDS). Indeed, before finding a refined P4 model "E" Prescott, I found the Celeron D (Prescott) 2.4 preferable to the Northwood for the short term since it was a fraction of the price and performed virtually the same when not multi-tasking (since it lacked HT).
Here are some old 3DM01 benchmarks with a Radeon "9700" (9500 256-bit 8pp). The first is at default viddy clocks while the rest are o'erclocked. The drivers vary but it still roughly demonstrates the differences. The main point being that the comparable Northwood simply could not compete with the Prescott on price and even if it could was not guaranteed to so effortlessly achieve the same performance (at default voltage no less). So, while technical arguments are fun they don't necessarily apply in practice.
P4 2.4
10561
P4 2.4 @ 2900
15834
C 2.4 @ 3060
15945
P4 3.0 @ 3014
16896
P4 3.0 @ 3608
18509
If you want the most indepth information on Prescott, here is the best article I have read on the subject.
http://www.anandtech.com/showdoc.aspx?i=1956
http://www.anandtech.com/showdoc.aspx?i=1956
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 + WCL Discussion Threads
- Started by Tigerick
- Replies: 22K
-
News NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products- Started by poke01
- Replies: 366
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
Facebook
Twitter