##testing## pci-e 2.0 X16 vs. pci-e 2.0 X8

[philbrown23]

as this is a common arguement I intend to prove that there is not much difference at al.

first off vantage: no real difference only like 20 points.

[Laurijan]

Really helps me think more positive about 8x SLI and Crossfire mobos..

post some game results too

[TRIPTEX_CAN]

Im interested in 2 x 4850 and this is the type of thing I was looking for. I would really like to see some game results too!!

ps: nice desktop

[FR@NK]

Try using a higher resolution....like the extreme performance test in vantage.

[Fitseries3]

i need to see if it makes a difference with monster cards like my 2 4870x2s.

[lemonadesoda]

If there is little difference between PCIe x16 and PCIe x8, then that is also confirmation that AGP is still up there (Since PCIe x8 and AGP are essentially the same 1way bandwidth). This is not to say that AGP is better. No PCIe is a great step forward, esp. as you can run many channels in total on a mainboard. BUT, AGP is still viable as a gaming platform.

RETRACTION

I see you are referring to PCIe v2.00 that has double the bandwidth of PCIe.

AGP is behind.

[Kwod]

I think Aussie website Tweaktown did a test with 8 and 16x, and the 16x slaughtered the 8x.
I'm off to sleep now, but you can google it and learn that Crysis will suffer without 16x2.

[TRIPTEX_CAN]

Originally Posted by Kwod

I think Aussie website Tweaktown did a test with 8 and 16x, and the 16x slaughtered the 8x.
I'm off to sleep now, but you can google it and learn that Crysis will suffer without 16x2.

The thing is.. Everyone is simply accepting the Tweaktown review as fact and nobody else has tried to prove/disprove them. It is possible that the P45 sample they used was simply not the best performer. Also in the comparison they made no mention of the p45 smoking the x48 is several of the benchmarks @ 1280x1024. While they are quick to point out how awful the P45 performs @ 2500x1600 with full AA and AF. The tests @ 1900x1200 showed the differences was much smaller. Im not saying Tweaktown did a poor comparison.. I would just like to see someone else run the tests.

Crysis will suffer no matter what system you throw at it. If a 9800gx2 can run with the 4870x2 the game must have issues beyond the hardware used to render it.

[philbrown23]

well my max resolution is 1440X900 but anyone who has the ability to use X16 or X8 is welcome to add reluts and they will go in the first page, I know about the tweaktown article, I have to say though...bullshit.

[TRIPTEX_CAN]

Originally Posted by fitseries3

i need to see if it makes a difference with monster cards like my 2 4870x2s.

fit with all the hardware you have just in your specs you could run the same tests as Tweaktown.

[philbrown23]

rofl fit post up some comparison shots at X8 and X16 with it.

[Sonido]

What everyone seems to forget is that a PCI-Ex 2.0 x8 is essentially a PCI-Ex 1.0 x16, so the difference isn't going to be monstrous; it is noticeable though.

[zithe]

Originally Posted by lemonadesoda

If there is little difference between PCIe x16 and PCIe x8, then that is also confirmation that AGP is still up there (Since PCIe x8 and AGP are essentially the same 1way bandwidth). This is not to say that AGP is better. No PCIe is a great step forward, esp. as you can run many channels in total on a mainboard. BUT, AGP is still viable as a gaming platform.

RETRACTION

I see you are referring to PCIe v2.00 that has double the bandwidth of PCIe.

AGP is behind.

AGP X8 = PCIe 2.0 x4

[Kwod]

Originally Posted by TRIPTEX_MTL

The thing is.. Everyone is simply accepting the Tweaktown review as fact and nobody else has tried to prove/disprove them. It is possible that the P45 sample they used was simply not the best performer. Also in the comparison they made no mention of the p45 smoking the x48 is several of the benchmarks @ 1280x1024. While they are quick to point out how awful the P45 performs @ 2500x1600 with full AA and AF. The tests @ 1900x1200 showed the differences was much smaller. Im not saying Tweaktown did a poor comparison.. I would just like to see someone else run the tests.

Crysis will suffer no matter what system you throw at it. If a 9800gx2 can run with the 4870x2 the game must have issues beyond the hardware used to render it.

IMO, I don't think a website reviewer can get away with such a "big lie"......other websites would quickly contradict Tweaktown and make a name for themselves whilst ridding some of their competition.

I think the P45's are good mobo's, I haven't read anything bad about them, but if you're going to XF then presumably you want to play at 1920x1200 or 1600p.
Based on what I've read, demanding games and future games will want 2x16 and 1 gig cards for maximum performance.

ASFAIK, there's no penalty when using a 4870x2 in a P45.

[Mistral]

ASFAIK, there's no penalty when using a 4870x2 in a P45.

There's no penalty using a 4870x2 in a nForce4 either, safe for the not-so-fast CPU.

That being said, I did order a P45 board, but I doubt I ever go CrossFire.

[TRIPTEX_CAN]

Originally Posted by Kwod

IMO, I don't think a website reviewer can get away with such a "big lie"......other websites would quickly contradict Tweaktown and make a name for themselves whilst ridding some of their competition.

I think the P45's are good mobo's, I haven't read anything bad about them, but if you're going to XF then presumably you want to play at 1920x1200 or 1600p.
Based on what I've read, demanding games and future games will want 2x16 and 1 gig cards for maximum performance.

ASFAIK, there's no penalty when using a 4870x2 in a P45.

I dont think Tweaktown lied... I just think they were very quick to point out how slow the p45 is.. There is also a chance that the p45 board they used simply needed work and didnt handle the DDR3 bandwidth very well. Also they did say the x48 board is only %10 more than the p45 which is simply rediculous since most p45 board are close to $150 and the x48 board are all close to $300. They just happened to use one of the most expensive p45 board available and in that case it would be smarted to use the x48.

No there should be no penalty running a 4870x2 in a p45 board.

[burebista]

Originally Posted by TRIPTEX_MTL

I would really like to see some game results too!!

User test on ocforums with 2x HD4870
Review on Legion Hardware with 2x HD4850.

[DaMulta]

even if you run CF or SLi in 8x you will not see a difference.

The only cards that I think you might is the x2. And maybe only runing that in CF or SLi mode.

[TRIPTEX_CAN]

Originally Posted by burebista

User test on ocforums with 2x HD4870
Review on Legion Hardware with 2x HD4850.

The first test looks much better for the p45 but ouch that p35 board is not looking so good. The second test confirms my thoughts. Either Tweaktown's p45 was simply slow or it didnt handle the DDR3 very well.

thanks again

[Tatty_One]

Originally Posted by Sonido

What everyone seems to forget is that a PCI-Ex 2.0 x8 is essentially a PCI-Ex 1.0 x16, so the difference isn't going to be monstrous; it is noticeable though.

Thats true in part but PCI-E 2.0 is more than double the bandwidth of 1.0, unless you meant 1.1 which is 2.5Gbit. The thing is, as said in the Op, there is little difference but as the resolution, and more importantly, the "power" of the card tested increases then the gap becomes far more noticeable, for example, the first graphics card to exceed AGP 8x specification was the 8800GTX/Ultra, however, PCI-E 1.0 was available 1-2 years before those cards were on the shelves (my guesstimate there).

The new top dog....the HD4870x2 is pushing greater than the 2.5Gbit 1.1 spec bandwidth so the differences in performance between 1.1 and 2.0 OR 8x and 16x will be more prominent...........if that makes any sense?

Here is an extract from the Wiki.....................

In PCIe 1.1 (currently the most common version) each lane sends information at a rate of 250 MB/s (250 million bytes per second) in each direction.
PCIe 2.0 doubles this data rate, introduced in late 2007, PCIe 2.0 is found on newer systems such as those based around the Intel X38 or AMD 780G chipsets.
The latest proposed PCIe 3.0 standard will increase the speed of the links further (tentatively scheduled for release around 2010).[1]

Each PCIe slot carries either one, two, four, eight, sixteen or thirty-two lanes of data between the motherboard and the addin card. Lane counts are written with an "x" prefix e.g. x1 for a single-lane card and x16 for a sixteen-lane card. Thirty-two lanes of 250 MB/s (PCIe 1.1) gives a maximum transfer rate of 8 GB/s (250 MB/s x 32, i.e., 8 billion bytes per second) in each direction. However the largest size in common use for PCIe 1.1 is x16, giving a transfer rate of 4 GB/s (250 MB/s x 16) in each direction. Putting this into perspective, a single lane for PCIe 1.1 has nearly twice the data rate of normal PCI, a four-lane slot has a transfer rate comparable to the fastest version of the old parallel PCI-X 1.0, and an eight-lane slot has a transfer rate comparable to the fastest version of AGP. However the data rates cited must be derated because 8b/10b encoding is used in the physical layer. The link transfer speeds cited are to be considered maximum theoretical data rates.

PCIe slots come in a variety of physically different sizes referred to by the maximum lane count they support, ie. x1, x2, x4, x8, x16 and x32. A PCIe card will fit into a slot of its size or bigger, but not into a smaller PCIe slot.

[SystemViper]

also Foxconn is comming out with the Avenger a P45 board that uses a cross;inx chip and runs at 2 x 16x 2.0

[TRIPTEX_CAN]

I wonder if its going to OC well. I considered a foxconn x48 board until I found out it probably wouldn't OC well.



The copper heatsync between the PCI-E slots is covering the chip you mentioned.

[SystemViper]

i head it' susposed to be great, i just hope is doesn't costan arm and a leg

[Tatty_One]

Originally Posted by TRIPTEX_MTL

I wonder if its going to OC well. I considered a foxconn x48 board until I found out it probably wouldn't OC well.



The copper heatsync between the PCI-E slots is covering the chip you mentioned.

A couple of their top end "special edition" boards are supposed to be monsters.....but pretty pricey though.