SiS648 vs. i845E/G

[TalonFyre]

Which has better memory bandwidth? Which has a better memory controller? I've seen some preview benchmarks of the new SiS648 using DDR400 which is near PC1066 performance on an i850(E) system. But they don't compare it too the results of i845E/G using DDR400. They only test the i845's with their default memory types (DDR266 / DDR333). I realize that nobody really has an SiS648 board yet but assuming that its memory performance/bandwidth is similar - if not slightly better - than the SiS645DX, does the i845E/G come close when running DDR400?

As a question-on-the-side, does anybody know if the new Asus SiS648 board has AGP/PCI bus lock feature? And is this feature chipset-specific or manufacturer-implemented?

 

[Athlon4all]

845E sucks bad. No DDR333 support, and its memory controller is slightly worse that 845g. The bottom line is that the 845e is 5-15% slower than 845g with DDR333. That being said, 648 has better performance than both of them. The 845g is 5-10% slower than 850e with PC1066, whereas the 648 is >5% slower than 850e+PC1066 in all but 2 benchmarks (Both SPECViewperf ones). Go 648

 

[OneOfTheseDays]

648

 

[Wingznut]

Personally, I wouldn't even begin to recommend the SiS 648 board yet. Even though the initial numbers look fairly impressive, we know absolutely nothing about the stability/reliability of the board.

A good benchmark will do you only so much if the platform has stability issues, or memory compatibility issues (such as Via's P4X400 chipset.)

 

[OldScratch]

I may be able to provide some information on the SIS648 in a couple of days. I just ordered the ABIT Model SR7-8X (SIS648) from newegg. It is still in stock as of this posting.

 

[OneOfTheseDays]

umm conisdering how the 648 is basically an extension off of the 645DX chipset I can assume that stability will be top-notch like the 645DX chipset was. And conisdering how all major mobo manufacturers are producing mobos based on the 648 chipset i think we can assure that it will be stable, after all why would they all make a mobo if it was inherently unstable?

 

[Wingznut]

My SiS645 board had quite a few issues. So, if the 648 hasn't improved much in that area, then I certainly cannot recommend it at all.


"after all why would they all make a mobo if it was inherently unstable?"

Heh, you act like that has never happened before.

 

[Budman]

why would they all make a mobo if it was inherently unstable?

(cough) VIA (cough) ..

 

[WangButter]

Originally posted by: Wingznut PEZ
My SiS645 board had quite a few issues. So, if the 648 hasn't improved much in that area, then I certainly cannot recommend it at all.

Which 645 board did you have? It may have had more to do with the board manufacturer than the actual chipset.

The main thing that hurt the 645 and 645DX is the lack of PCI / AGP lock. I'd be interested to see if the 648 still suffers from this issue.

 

[Wingznut]

Originally posted by: WangButter

Originally posted by: Wingznut PEZ
My SiS645 board had quite a few issues. So, if the 648 hasn't improved much in that area, then I certainly cannot recommend it at all.

Which 645 board did you have? It may have had more to do with the board manufacturer than the actual chipset.

That's true... it was an MSI.

 

[OneOfTheseDays]

well wingnut you obvious didn't do your homework. The MSI board is based off of the 645 chipset, not the 645DX chipset. The best mobo based off of the 645DX is the Asus P4s533, it's only lacking is the lack of the PCI/AGP lock in the bios. If the new 648's don't have that feature, i'm not upgrading.

 

[Wingznut]

So, you're saying that the AGP driver issues that I had with the 645 aren't present in the 645DX, even though they use the same driver (as does the 648)?

Heh, looks like I did alright on my homework.

But before we get into nitpicking the issues I had (as I've seen you go around and around with other members before), let's just remember my initial point.

It's much too early in the game to wholeheartedly recommend a board based on new chipset. It happens around here very often... A new board will be released, and everyone will fawn all over it because of the good benchmark results. Five weeks later, everyone is jumping off the bandwagon because of stability/compatibility issues.

 

[Mikki]

Originally posted by: Athlon4all
845E sucks bad. No DDR333 support, and its memory controller is slightly worse that 845g. The bottom line is that the 845e is 5-15% slower than 845g with DDR333. That being said, 648 has better performance than both of them. The 845g is 5-10% slower than 850e with PC1066, whereas the 648 is >5% slower than 850e+PC1066 in all but 2 benchmarks (Both SPECViewperf ones). Go 648

845E does NOT suck bad. Or even a little bit, and I don't understand how you can say that. Actually, at least with the P4B533-E that I have experience with, it is a pretty awesome for a board with all the goodies on it (raid, usb 2, sound, etc). No, it doesn't have 333 support (P4B533-E, 1.6a @ 2.4, ddr400, and very satisfied), and maybe it doesn't perform as well (so you're telling me you're able to notice a 5% performance difference? come on!) as others. And it's cool how you compare the 845 to the 850 with 1066 , yeah that's fair. And as long as we're on the subject, from what I've seen in these and other forums, the SiS chipsets DO have issues that Intel chipsets don't. They perform well and I consider them everytime they come out, but then I see people's problems and.........