4200+ vs 4400+

[ab]

Hi,
I want one of these. Does anyone know which o/c's higher on average?
Does the extra cache on the 4400+ present problems like it used to with
the (very old) P2/celeron line.

Tom

 

[General Schvantzkoph]

Hi,
I want one of these. Does anyone know which o/c's higher on average?
Does the extra cache on the 4400+ present problems like it used to with
the (very old) P2/celeron line.

Tom

The extra cache significantly improves performance. Here are some
benchmarks that I ran on my systems. The 3800+ has a faster clock but only
1/2M of cache. The 3400+ and 4400+ both have 1M caches. The two benchmarks
are a NCverilog, No IO writes no data, Hi IO writes a large data file. On
the no IO case the 4400+ is twice as fast as the 3800+, when there is a
lot of IO then the main memory bandwidth (which is the same for both the
3800+ and the 4400+) dominates.

Machine CPU NCVerilog NCVerilog Relative to Wasp
No IO Hi IO
Wasp 3800+ 56.48 204.65 1 1
Ranger 3400+ 35.3 195.03 1.6 1.04
Nimitz 4400+ 27.78 179.7 2.03 1.13

My 4400+ has very little overclocking capability. I was able to up the
clock 5% and still boot the OS. I tried 15% and it wouldn't even boot into
the BIOS, had to use the CMOS clear button to fix the system.

You could get lucky but I wouldn't count on overclocking either a 4200+ or
4400+, they are both brand new so there isn't going to be a lot of margin
in the process. If you are looking for the fastest system then get the
4400+, the difference in price is small and there is a huge performance
gain in some cases.

One more thing, don't get an MSI motherboard. The MSI's have a bug where
they only recognize 3.5G out of 4. Both the K8N Neo2 and K8N Neo4
behave this way (I have both). Someone else reported that their Abit sees
all 4G, can't confirm that this is true but the MSIs are definitely not
the way to go.

 

[Mark W. Smith]

The extra cache significantly improves performance. Here are some
benchmarks that I ran on my systems. The 3800+ has a faster clock but only
1/2M of cache. The 3400+ and 4400+ both have 1M caches. The two benchmarks
are a NCverilog, No IO writes no data, Hi IO writes a large data file. On
the no IO case the 4400+ is twice as fast as the 3800+, when there is a
lot of IO then the main memory bandwidth (which is the same for both the
3800+ and the 4400+) dominates.

Machine CPU NCVerilog NCVerilog Relative to Wasp
No IO Hi IO
Wasp 3800+ 56.48 204.65 1 1
Ranger 3400+ 35.3 195.03 1.6 1.04
Nimitz 4400+ 27.78 179.7 2.03 1.13

My 4400+ has very little overclocking capability. I was able to up the
clock 5% and still boot the OS. I tried 15% and it wouldn't even boot into
the BIOS, had to use the CMOS clear button to fix the system.

You could get lucky but I wouldn't count on overclocking either a 4200+ or
4400+, they are both brand new so there isn't going to be a lot of margin
in the process. If you are looking for the fastest system then get the
4400+, the difference in price is small and there is a huge performance
gain in some cases.

One more thing, don't get an MSI motherboard. The MSI's have a bug where
they only recognize 3.5G out of 4. Both the K8N Neo2 and K8N Neo4
behave this way (I have both). Someone else reported that their Abit sees
all 4G, can't confirm that this is true but the MSIs are definitely not
the way to go.

I can confirm that my Abit does. CPU-Z pic is on the web site below.
I had to update the bios for core #2.



--
http://www.mark-w-smith.net/cpu-z.htm


Athlon64 4400+
4GB PC3200
Abit Fatal1ty nForce4 SLI
NVIDIA GeForce 6800




---
avast! Antivirus: Outbound message clean.
Virus Database (VPS): 0534-4, 08/26/2005
Tested on: 8/27/2005 6:42:34 PM
avast! - copyright (c) 1988-2005 ALWIL Software.
http://www.avast.com

 

[General Schvantzkoph]

I can confirm that my Abit does. CPU-Z pic is on the web site below.
I had to update the bios for core #2.

I managed to get it working on the MSI, added mem=5g to the kernel line in
/etc/grub.conf.