PCI-E 2.0 vs 1.0 impact on performance?

Just check any semi-recent slide from nV or ATI/AMD, they always start at 80%

 

Thats whaty i'm talking about, that is wy i am selling hte blitz extreme and just purchased the macxumus extreme, true 16 x 16 x crossfire and pcie2, but now i got to figure out how to watercool, have a great holiday

Regards
Rhinoviper

 

You have 3 bus controllers here with 3 different architectures and it's obvious that they will perform differently in different games.
But that has nothing to do with PCI-E bandwidth.
The assumption that the bandwidth is the only thing that's different in these chipsets is kinda wild (or should I say: plain wrong?).

 

Titles? Do you mean that you haven't done any synthetic tests?

 

2 boards are P35, 1 is X38.

 

oh gee, i guess i am just plain wrong

 

One of the P35 uses southbridge to get second PCI-E data.
Layout is different and latencies are different -> different performance.

 

Ah, thanks for pointing that out.

Speaking of P35, can anyone recommend a good board for CF that will allow for good CPU overclocking? I'm thinking the Abit IP35 Pro is the board I want, but I also like Gigabyte's boards. Wouldn't be opposed to an Asus board, but they're usually way overpriced.

 

Abit has good reputation among P35 boards, but Gigabyte has been getting good words out there too. ASUS has apparently some issues.

 

Yeah, looks like Gigabyte or Abit for me. Unless someone comes out with a decent X38 alternative that's not significantly more expensive than what the IP35 Pro is already going for.

 

Amazing then that reviews rarely show differences greater than 1-2% despite wildly different MB layouts and configurations.

Although I suppose it is possible in some weird contrived test to have one get up to a 20% difference...although it boggles the mind to think what that would be.

http://www.anandtech.com/mb/showdoc.aspx?i=3149&p=6

For example contains boards from DDR 2 to DDR 3. And Chipsets from the 965, 975 to p35.

And other than the Intel board that is tuned for stability and NOT performance, amazingly enough they are all within (gasp) 1-2% of each other.

So what was that again about wildly different performances in different games based on the chipset or hell even the memory used?

Even if we were to be generous and say a p965 using DDR2 memory will run lets say 5% slower (you'll be hard pressed to actually find a game that displays this behavior on performance/enthusiast class MBs) that still doesn't account for the up to 49% difference in that article.

Hell only the Fear benches from that site would fall into the margin of error and possible differences in chipsets.

Although the results for WIC are odd to say the least.

Regards,
SB

 

i am not sure if i am in the right place. this is my computer specs:
Feature Specification
Form Factor MicroATX 244 mm × 244 mm
Processor Support Supports AMD® Socket AM2+ (940 pin), AMD Phenom™ (95 W), Athlon™ 64 X2, Athlon 64, and Sempron™ processors
Chipset Northbridge: RS780
Southbridge: SB700

System Bus Support HT3 / HT1
Memory Support Four DDR2 800/667 MHz slots support up to 8 GB system memory
Video Integrated video

Audio Integrated ALC888S HD codec 7.1
Network Support Integrated Marvell 8071 / 8075 10/100/1000 Mbps LAN
Add-In Card Slots One PCI Express ×16
One PCI Express ×1
Two PCI conventional
it has a nvidia geforce 6150SE nForce 430 and i was wanting to upgrade to the ATI radeon hd 4600 series. is this possible and if not what is my best option. thanks folks.

 

sounds like you have a small form factor case
2 worries
will it fit
is your psu up to the job

 

Its not needed IMO, i run an 8800GT SLI set-up on my rig which is running on an ASUS M2N-SLI Deluxe mobo that is PCI-E 1.0.

I wondered about his when i brought the motherboard as i was scared that it would hinder and hold back my GT's. So when i got the board about 2 months ago i tested a single GT in the x16 and x8 slot and run some benchmarks to see if there was a bottleneck present. I tried all sorts of games, Crysis, bioshock, counterstrike, doom 3, Quake 4 Oblivion, STALKER and no matter what slot i used the card performed the same in each slot.

Anandtech did an article on this and they also found that there is no difference.

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3180&p=5

 

There's benchmarks from few sites, though, which show clear difference on performance between PCIe 1.x and 2.0 when using HD4800-series cards in CF

 

did we ever come close to saturating Agp8x ?

 

Well from my personal experience theres no difference between PCI-E2.0 & 1.0 in terms of performance and im running SLI.

Ive also seen reviews were a 280GTX was tested in a PCI-E 1.0 slot and it performed the same as it did in its native 2.0 slot.

 

Here, let me help:

Crossfire scaling with a pair of 512mb 4850's on a P35 vs P45 vs X38 vs X48. In several cases, they nuked the extra lanes via software config. Looks to me like if you want good crossfire scaling, PCIE 2.0 provides far more than single-digit increases on apps that aren't video limited, but you don't necessarily need the full 16x...

 

Here some PCI-E ver.1.1 VS. 2.0
http://www.tomshardware.com/reviews/crossfire-pci-express,2095.html