Intel C2D vs A64

[Compman55]

Back in the day before core i3/i5/i7 was ever created, which was a better CPU the Intel C2D or the amd Socket 939 series of cpus (A64, Opteron, FX, etc....)

 

[masteryoda34]

Core 2

 

[Plimogz]

just to pile on and put this to rest: C2D destroys A64. And 939 are really getting on in years at this point; motherboard failure cannot help but be an issue when dealing with old 939 boards.

 

[SPBHM]

AMD was very competitive from 1999 (athlon) up until 2006 when Core 2 Duo was released,
basically Core 2 Duo was far superior from what AMD had (Athlon 64 X2), but in 2003-2005, AMD normally had faster, and more power efficient CPUs.

 

[tweakboy]

Core 2 all the way!

 

[Red Hawk]

http://www.anandtech.com/show/2045

As SPBHM said, AMD was quite competitive, and at times decisively better, than Intel up until Intel's first wave of Core 2 Duo processors (Conroe architecture). Core 2 Duo came along and blew AMD's lineup out of the water -- as Anand's review notes, Core 2 Duo didn't lose a single benchmark, and more often than not had the ~$350 C2D E6600 processor matching up with AMD's $1000 processor. AMD has never recovered since that loss.

 

[tweakboy]

X2 wont come near it as well..

 

[pantsaregood]

I think Core 2 vs. Athlon 64 X2 is often misrepresented. Everyone knows that K8 was superior to NetBurst, but people fail to realize that by the end of Pentium D's life as a "premium" product, AMD was completely smacking Intel around in performance. The performance gap between K8 and NetBurst was greater than the performance gap between Conroe and K8.

 

[Arachnotronic]

Pentium 4/Netburst was absolute garbage. I remember owning a Pentium D 805...awful chip, but it overclocked well. Pentium D 925 was better, but my Q6600 was a godsend. Wish I had been using AMD throughout those years, but I was still bitter about the chipset issues I had in the Athlon XP days (damn you, Via KT-333).

Pentium 4 "Northwood" was good...but then they screwed it up with Prescott. Dumb move. Intel was such a mess in those days...it's amazing that they actually survived that.

 

[2is]

I've had a few 939 X2's as well a AM2 X2's and Core 2 is by FAR a better processor, and clocks better more often than not also.

 

[Phynaz]

Core2 is the CPU that was too good. It killed AMD.

 

[dma0991]

Had a AMD Athlon 64 3800+ desktop and a C2D laptop back then. Its an unfair comparison of a single core vs dual core but the desktop CPU has nearly double the clockspeed.

The laptop is still very usable even today but the desktop struggled with watching HD movies, transferring files, etc. All I got was stuttering video playback and 100% CPU usage just to transfer files.

 

[2is]

Kind of irrelevant really. Core 2 Duo kicked the crap out of X2's as well, so either way you look at it, Core 2 is the better CPU

 

[pantsaregood]

Pentium 4/Netburst was absolute garbage. I remember owning a Pentium D 805...awful chip, but it overclocked well. Pentium D 925 was better, but my Q6600 was a godsend. Wish I had been using AMD throughout those years, but I was still bitter about the chipset issues I had in the Athlon XP days (damn you, Via KT-333).

Pentium 4 "Northwood" was good...but then they screwed it up with Prescott. Dumb move. Intel was such a mess in those days...it's amazing that they actually survived that.

Prescott is remembered being worse than it actually was. Per-clock performance was a toss up (and by shallow margins) between Prescott and Northwood, but Prescott introduced support for some newer instructions. Prescott also pushed NetBurst to higher speeds as Intel intended.

Prescott likely could've scaled further than it did. My grandmother's 3.6 GHz Celeron D (actually a Cedar Mill) runs at 4.5 GHz without any stability problems - the only reason it isn't pushed further is because the power supply can't handle it. Power consumption, even though it dropped with Cedar Mill, was becoming too much to deal with. Intel could've pushed out models up to 4.2 or 4.4 GHz most likely, but they would've remained unable to compete with AMD and they would also ultimately face the same problem.

This is a rather odd relic of Intel's original plans:
http://ark.intel.com/products/27510...-HT-Technology-4_00-GHz-2M-Cache-1066-MHz-FSB

 

[Hulk]

The performance gap between K8 and NetBurst was greater than the performance gap between Conroe and K8.

I respectfully disagree with the above comment. Here is the Anandtech initial review of the C2D, pitting the fastest Athlon, C2D, and Pentium 4 of the day against each other.
http://www.anandtech.com/show/2045/8

The Athlon was faster than the P4 but if you look at all of the tests, not as much as the delta between the Athlon and C2D. For example,

Sysmark 2004 (overall)
C2D - 371
Athlon FX-62 - 273
Pentium EE965 - 256

Of course this is just one example and to be fair there are examples of the FX62 evenly spitting the C2D and 965 performance. But overall the FX62 was closer in performance to the 965. The C2D was in another league.

 

[inf64]

C2D had around 20% IPC advantage,on average, over K8 (90nm,1MB L2). This is by no means "destruction" but is indeed solid advantage. What really made Conroe noticeably better desktop chip is that it OCed very good (vs limited results on K8) on top of that IPC advantage.

Now when Phenom(I) launched it almost nullified the IPC advantage intel had (brought it down to ~6% on average), but did nothing on the clock nor power front. This can be seen here. Relative performance of Phenom X4 9750 @ 2.4Ghz, if selected on the chart, was 100 and rel. performance of Q6600 @ 2.4Ghz was 106.3. The fact was C2D/C2Q still clocked much better and it wasn't until B3 stepping of 65nm Phenom when AMD finally came a bit closer, but still not nearly on par with OCed C2Qs. Then Phenom II launched, brought clock(stock/OC) and IPC gains and it was very close to 45nm Penryn parts but AMD's problem was that it was now going against much improved Core, Nehalem. So from then they are playing catchup game with a competitor that has much stronger core, much better manufacturing process and tons of money to burn on RnD. They made a slight step back with BD core but are now getting back on track with PD and hopefully SR, which is their last chance to come back in the performance game.

 

[aaksheytalwar]

The slowest core 2 beats the fastest x2 is modern stuff I guess. A64 is single core and was a competitor to Prescott but a bit better for games than Prescott.

 

[24601]

The Core 2 Duo was so good Intel got away with using Thermal Interface Material between the IHS and the CPU Die instead of direct soldering on many of the early Core 2 Duos (And my very own E4400 2.0 ghz @ 3.5 ghz).

Intel was deliberately limiting the performance of their Core 2 Duos in the beginning to have the longevity it has had since. This is why they overclocked so ungodly well.

 

[Lorne]

The C2D cpu crunched better with only its chipset limiting its bandwidth as it still used the off NB MMU, But OCing that helped.
AMD's on core MMU is the only thing that helped them hold the line till the I-series came out.

The only thing the P4 improved over the P3 was on chip cache and HT, Other then that it was a step backwards.

 

[Compman55]

Pentium 4/Netburst was absolute garbage. I remember owning a Pentium D 805...awful chip, but it overclocked well. Pentium D 925 was better, but my Q6600 was a godsend. Wish I had been using AMD throughout those years, but I was still bitter about the chipset issues I had in the Athlon XP days (damn you, Via KT-333).

Pentium 4 "Northwood" was good...but then they screwed it up with Prescott. Dumb move. Intel was such a mess in those days...it's amazing that they actually survived that.

Intel17,

What kind of chipset issues did you have with the KT333? I had a gigabyte GA-7VRXP in which the via hyperion 4.43 solved every issue. And later the 4.49 version became the norm on all my builds. I had sound crackle, and choppy video until this driver came out. Athlon XP days were the best. Then ICH5R came along with awsome RAID perforance and that ended my AMD days.

 

[pantsaregood]

The slowest core 2 beats the fastest x2 is modern stuff I guess. A64 is single core and was a competitor to Prescott but a bit better for games than Prescott.

The slowest Core 2 Duo certainly doesn't beat the fastest Athlon 64 X2. Core 2 wasn't the ridiculous leap forward people remember it as. Pentium 4 was just behind.

Take a look:
http://www.anandtech.com/show/2056/6

Not many people remember Yonah, but it was essentially an enhanced Pentium M. You'll notice that, while Merom certainly improved on Yonah in IPC, it didn't destroy it in the sense that people remember.

Conroe is remembered as such a game changer because of this:
http://www.cpu-world.com/Compare/88...ium_Extreme_Edition_965_(HH80553PH1094M).html

 

[jacktesterson]

I still remember when I just built my $2000.00 T-Bird 1.2GHz gaming system with a Geforce 2 Pro

I was ballin

Of course, not as ballin as my $3000 Pentium 166 MMX that could barely do anything a year later when PII/Celerons came along.

 

[RampantAndroid]

Pentium 4 "Northwood" was good...but then they screwed it up with Prescott. Dumb move. Intel was such a mess in those days...it's amazing that they actually survived that.

We had a box at work with a prescott core in it. We named the machine spaceheater. It's shared files were at \\spaceheater. If the office was cold, fire off a build...

Anyway, I think an Intel engineer came out and said Netburst contributed to their current success, since getting Prescott stable at speed took so much effort they learned a lot about design that they are still benefiting from.

 

[Cerb]

Not many people remember Yonah, but it was essentially an enhanced Pentium M. You'll notice that, while Merom certainly improved on Yonah in IPC, it didn't destroy it in the sense that people remember.

Yonah only existed in an affordable format in Mac Minis, some other SFF PCs, and embedded computers. Yes, you can follow the history of the P6 as a series of gradual improvements, but on the desktop, we had Pentium 4 CPUs, Pentium D CPUs, and Athlon64 X2 CPUs, offering decent performance.

The Pentium D CPUs could be slower than regular Pentium 4 CPUs w/ HT. The P4 and P D CPUs generally were slower than Athlon64 CPUs, and were much hotter, a serious issue for non-U.S. buyers, by that time. Also, the P4EE CPUs (Xeons named Pentium) were quite expensive (AMD was supply-limited). Athlon FX CPUs were expensive, too, but you sacrificed almost nothing getting a regular A64 X2, unlike with the P4s. Like A64 X2s, P4EEs still offer acceptable desktop performance, today (I recently upgraded one to Win7), much more-so than P4 or P D CPUs (but oh, do they run hot!).

The Core 2 Duos came out cheaper than P4EE or Athlon FX CPUs, while performing as well or better, and offering better average power consumption.

 

[pantsaregood]

Pentium 4 EE didn't really offer anything over the standard models. It had 2 MB of L3 cache, but that didn't really do much to improve performance. Those models topped out at 3.46 GHz. If memory serves correctly, the entire reason P4EE came into existence was to immediately combat the release of K8.