Is loss of Dual Channel in upgrade significant?

[cee jaay]

When I upgraded from a Gigabyte 7N400 L1 (nforce2) with AthlonXP 2600 and 2*256 DDR - Dual Ch to a Gigabyte 8NSNXP (nForce3) with Athlon64 3200 and 1*512 DDR - Single Ch. The increase in speed was significantly lower than I'd expected (3dMark even gave a lower result after the upgrade). Unfortunately, I hadn't realised the 754 socket boards were only single channel.

My question is would the lack of "Dual Channel"ing on the new mobo likely to be causing this?

Some have said that this feature delivers approx 5-10% improvement and I'm wondering whether getting 2*512Mb of cracking top-end Corsair memory would be a better and more cost effective option than replacing the mobo and chip to 939s.

 

[TheInvincibleMustard]

Yes, dual-channel does provide a modest improvement over single-channel, but it's not as needed on K8 architecture as it is on the long-pipelined P4 architecture. Going from dual-channel to single-channel and changing your platform/cpu might just about even things out in that regards.

RE: lower speeds -- did you perform a fresh install of the operating system when you upgraded? It's not necessary, but there could be lingering junk around if you didn't. I'm also curious as to what your memory settings are, if you're running the 512 at DDR400 or what. There isn't as much of a performance difference betweenmemory speeds on the A64 side as on the P4 side, due to the way the memory controller is set up, but it's a point to bring up regardless.

Also, those expensive Corsair modules aren't all they're cracked up to be. Take a gander over at this forum post and see for yourself. Something like these Mushkin sticks are just as good for the A64 architecture.

HTH

 

[cee jaay]

Hi,

Yes, I always do a clean install for the reasons you mention (with a reformat of the XP partition).

I'm pretty sure the memory is running at 400Mhz. I've not tweaked them, so they'd be default. I've just started reading-up about latency settings and wondered if these are commonly adjusted. Thanks for the links.

(btw I did notice the CPU clock speed showing 200, which surprised me at first, but have guessed this is due to an Athlon64 running in a 32bit 'mode').

Thanks,
CJ.

 

[ChineseDemocracyGNR]

There's definately something wrong with your new setup. In this test an Athlon 64 3200+ (single channel) beats the Athlon XP 3200+ by 13% in 3DMark01 SE.

The Athlon 64 can operate normally in both 32 and 64-bit modes. The Gigabyte BIOS refers to what you're used to calling "FSB" as "CPU Frequency", but your processor should be operating at "CPU Frequency" * "K8 CPU Clock Ratio". To check your processor real clock run the "CPU-Z" utility and remember that if CnQ is enabled it will show it running at half its normal speed if your system is idle. Also use CPU-Z to check the memory clock.

 

[cee jaay]

Thanks for the pointers.

In CPU Z, the CPU page says
"AMD Athlon 64 3200+"
Core Speed: 2210.8Mhz
Multiplier: x11.0
HTT: 201Mhz

The Memory page says:
Frequency: 201.0Mhz
FSBRAM: CPU/11.0

Then on the SPD page I noticed:
Module Size: 512Mb
Max Bandwidth: PC3200 (200MHz)
Manufacturer: Hyundai

In 3dmark03, I'm getting 1914(!)
(btw the Graphics card's a Sapphire Atlantis 9800Pro Ultimate)

Looks like I've been sold slower memory and that's pegging the performance.

 

[mechBgon]

Originally posted by: cee jaay
Thanks for the pointers.

In CPU Z, the CPU page says
"AMD Athlon 64 3200+"
Core Speed: 2210.8Mhz
Multiplier: x11.0
HTT: 201Mhz

The Memory page says:
Frequency: 201.0Mhz
FSBRAM: CPU/11.0

Then on the SPD page I noticed:
Module Size: 512Mb
Max Bandwidth: PC3200 (200MHz)
Manufacturer: Hyundai

In 3dmark03, I'm getting 1914(!)
(btw the Graphics card's a Sapphire Atlantis 9800Pro Ultimate)

Looks like I've been sold slower memory and that's pegging the performance.

If your RAM is running at DDR400, then CPU-Z will show it at 200MHz since it's double-pumped.

Suggestion: move on from your 3DMark benchmarks to something real, such as UT2004 Demo Botmatch.

 

[SolMiester]

I would check for latest board drivers......

 

[cee jaay]

OK, I'm downloading the UT2004 demo as we speak.

And have grabbed the latest NVIDIA Windows nForce (Chipset) Drivers from Gigabyte (503).

One other thing does anyone know if slot #1 the best position for a single stick of memory (on this 3-slot board).

 

[darkswordsman17]

The slot closest to the CPU.

The newest drivers are 5.10.

Also get the newest video drivers.

But yeah, there's definitely something wrong.

 

[cee jaay]

Unfortunately the 290Mb download bombed-out last night.

flashed the bios - now on F10
am downloading 5.10 board drivers
and nbench if anyone is using that

If anyone has a GA-K8NSNXP (non-939) with 3200+ CPU, I'd dearly appreciate some stats for comparison

 

[cee jaay]

Okay, sytems specs with Botmatch are:

Operating System: Windows XP Home Edition (5.1, Build 2600) Service Pack 2 (2600.xpsp_sp2_rtm.040803-2158)
Processor: AMD Athlon(tm) 64 Processor 3200+, MMX, 3DNow, ~2.2GHz
Memory: 512MB RAM
DirectX Version: DirectX 9.0c (4.09.0000.0904)
Card name: RADEON 9800 PRO
Driver Version: 6.14.0010.6505


and the unreal tournament 2004 results are:

1024x768 64.31 fps botmatch
(dm-rankin: 77.61 fps, as-convoy: 47.45 fps, br-colossus: 67.88 fps )

Is this about right?

 

[cee jaay]

or not?