X48 vs P45 Crossfire Comparison... OUCH 8x/8x = fail

yea but P45 = $100 - $150 and X48 is $230+. Dunno about you but $259 is too much for a motherboard imo. Since I change them out so frequently I keep my motherboard purchases as close to $100 as I can.
 
P5E Deluxe is $220 and can be flashed to a Rampage Forumla. I was hoping to get the P5Q-E with all the neat features like Express Gate but oh well. Not sure how creditable that review from Tweaktown is as others disagree.
 
this is killing me. money is no issue and i got a p45, going to get a DFI X48 right after i sell off my POS p45
 
this is killing me. money is no issue and i got a p45, going to get a DFI X48 right after i sell off my POS p45

I was really liking the p5q lol, but looks like i'm giong with the dfi also, the DK x48 most ilkely, if not that then the gigabyte x48 ds4
 
I love how this was posted like 2 hours after i order. So with newegg I can just return and pay 15% restocking?
 
I love how this was posted like 2 hours after i order. So with newegg I can just return and pay 15% restocking?

They don't process orders on the weekend right? Go to newegg and see what stage it's in. You might still be able to cancel, then again, the credit card transactions might still take place on the weekend.
 
I placed my order this weekend and they charged the credit card quickly, so canceling is out of the question for me. I think I'll order an x48 board (probably a Gigabyte GA-X48-DS) then RMA for refund the P5Q Pro I will be getting this week ;/
 
I placed my order this weekend and they charged the credit card quickly, so canceling is out of the question for me. I think I'll order an x48 board (probably a Gigabyte GA-X48-DS) then RMA for refund the P5Q Pro I will be getting this week ;/

If it's too late to cancel but hasn't shipped yet you can try contacting Newegg to cancel it so you don't have to pay a fee to RMA it..
 
one review from a shady review site like tweaktown and you guys are ready to disregard p45 cf? get a grip.
 
If it's too late to cancel but hasn't shipped yet you can try contacting Newegg to cancel it so you don't have to pay a fee to RMA it..

That could be worth a try. The thing is, though, the P5Q Pro I ordered was part of a $30 discount combo deal with a 4850. You think I have a better chance of not having that discount taken away from me contacting them before it ships or doing the RMA? Or are they just too smart and will catch it either way?
 
one review from a shady review site like tweaktown and you guys are ready to disregard p45 cf? get a grip.
If I had already ordered the P5Q, I'd at least play with it and get your own benchmarks ready. I haven't ordered a mobo yet so I'm not going to axe the P45 just yet. A mass of numbers should hit the Internet this week to confirm or deny the results.
 
That could be worth a try. The thing is, though, the P5Q Pro I ordered was part of a $30 discount combo deal with a 4850. You think I have a better chance of not having that discount taken away from me contacting them before it ships or doing the RMA? Or are they just too smart and will catch it either way?

Well, if it doesn't cost more then $30 to RMA it, then I'd just do that because you'll still some mula out of the discount. I don't think they'll charge you an additional $30 for returning the mobo just because it was part of a combo deal.. But I'm not 100% sure either.
 
one review from a shady review site like tweaktown and you guys are ready to disregard p45 cf? get a grip.


Well first off, tweaktown has been around for a long time, and are typically a pretty reliable source. Second, simply put the p45 is supposed to be crippled due to the down scaling to 8x/8x with CF. Sure it might not have been as bad with the 3800's, but the newer cards have more bandwidth to move. So, while it's not normally a good idea to go off one review, I think we all know that it's probably pretty solid, knowing the source of the performance hit already.
 
one review from a shady review site like tweaktown and you guys are ready to disregard p45 cf? get a grip.

Please provide proof.

As I said in the other thread about this, Tweaktown is more excited for the HD4850 then us yanks because over in austraillia an HD4850 is going to sell for $250, whereas the 9800GTX is currently $350 minimum. Tweaktown and the majority of their readers have every reason to be excited for this card.
 
I was in the same situation as you guys were, thats why I created this thread after failure of getting a clear answer from asking questions. I have two HD4850's and needed to purchase a new mobo anyways because mine is AGP and a AMD. I was really close to ordered the P5Q-E but decided to wait because I wanted to spend a little bit more time reviewing it because I wanted to be sured that the x8/x8 crossfire wouldn't limit the cards.
 
One thing people are forgetting is that it's 8x/8x PCI-E 2.0. That is the same bandwidth as 16x/16x PCI-E 1.x. Considering earlier cards were barely bottlenecked by 8x/8x PCI-E 1.x, I highly doubt that 8x PCI-E 2.0 (= 16x PCI-E 1.x) is really that much of a hindrance to Crossfire. That said, I too will wait for more than one site's test, no matter how good that one site is (even the [H]), to come to any definitive conclusions.
 
I think I'm just going to get a 4780X2 and not deal with crossfire at all. I'm not sure I've very torn right now. On one hand I like the "newness" of the p45 and that it's ultra cool, but on the other I want to at least have the chance of running crossfire. BUT I ordered a Xonar D2X and it looks like it doesn't fit ANYWHERE. I'd have to use a 16 PCI-E slot.
 
One thing people are forgetting is that it's 8x/8x PCI-E 2.0. That is the same bandwidth as 16x/16x PCI-E 1.x.
I keep hearing this but I've never seen any proof of this and the specifications for PCIe 2.0 arn't based around huge performance jumps as much as development control of the GPU itself. Seems like a widespread rumor imo.
 
I think I'm just going to get a 4780X2 and not deal with crossfire at all. I'm not sure I've very torn right now. On one hand I like the "newness" of the p45 and that it's ultra cool, but on the other I want to at least have the chance of running crossfire. BUT I ordered a Xonar D2X and it looks like it doesn't fit ANYWHERE. I'd have to use a 16 PCI-E slot.
 
I keep hearing this but I've never seen any proof of this and the specifications for PCIe 2.0 arn't based around huge performance jumps as much as development control of the GPU itself. Seems like a widespread rumor imo.

The proof is on PCI-SIG's own FAQ. It's a doubling of bandwidth per PCI-E lane.

Of course you're not going to get huge performance jumps with PCI-E 2.0 when video cards aren't even using all of PCI-E 1.x's bandwidth. It's the same thing as SATA 3.0 Gb/s vis-a-vis SATA 1.5 Gb/s. The infrastructure is laid long, long before it's taken advantage of. That way, people have plenty of time to transition to new motherboards before new video cards (or hard drives) come out that really need the extra bandwidth.

For a single card, we're still some way from PCI-E 2.0 really being necessary. However, it's a happy coincidence that it comes just in time to keep 8x/8x motherboards viable for SLI/Crossfire.
 
A person at xtremesytems is going to be doing a test of 4850's CF on P45 tomorrow, I'd wait for their results before I make up my mind. If 4850 bottlenecks with an x8 PCI-E 2.0 then a 4870x2 should be bottlenecked by a x16.... The problem is the results are so far off from anything else you can find on the web, and also I have a problem with it only being certain games at certain reso's... I actually suspect drivers/bios/etc.
 
BW_Table.png



I will say those of you on P35 boards DO NOT go crossfire at all! the 4x slot really kills the performance.
 
I will say those of you on P35 boards DO NOT go crossfire at all! the 4x slot really kills the performance.

On P35, of course not. A 4x PCI-E 1.x slot (and P35 is PCI-E 1.1) has a quarter of the bandwidth of a PCI-E 2.0 8x slot (a la P45), or an eighth of a PCI-E 2.0 16x slot. P45 is a huge step up from P35 in the Crossfire arena.
 
The proof is on PCI-SIG's own FAQ. It's a doubling of bandwidth per PCI-E lane.
Thanks! I was under the impression there was little performance due to the fact PCIe 2.0 uses the same electrical contact methods, which it does, but the performance increase is from faster signaling rather than wider data paths. Had to dig pretty deep to find that one...
 
A person at xtremesytems is going to be doing a test of 4850's CF on P45 tomorrow, I'd wait for their results before I make up my mind. If 4850 bottlenecks with an x8 PCI-E 2.0 then a 4870x2 should be bottlenecked by a x16.... The problem is the results are so far off from anything else you can find on the web, and also I have a problem with it only being certain games at certain reso's... I actually suspect drivers/bios/etc.

I eagerly await the conclusions of this test.
 
He had problems with his board due to ram incompatibility, that's what he gets for using a DDR3 version... Still looking for other reviews :(
 
I gave in and went X48. If you want a single 4870, the P45 is the winner but not even going to take the chance as the X48 is more expensive for a reason...
 
Well, if it doesn't cost more then $30 to RMA it, then I'd just do that because you'll still some mula out of the discount. I don't think they'll charge you an additional $30 for returning the mobo just because it was part of a combo deal.. But I'm not 100% sure either.

From the Newegg website:
If you wish to request an RMA for an item you purchased as part of a combo or bundle special, you'll need to return all items within that particular combo / bundle special.

D'oh. Oh well. Anyway, I pulled the trigger on a P5E Deluxe, and just now, another ASUS 4850, since I have to return the P5Q PRO & 4850 combo in its entirety. This is becoming costly.
 
I gave in and went X48. If you want a single 4870, the P45 is the winner but not even going to take the chance as the X48 is more expensive for a reason...


You have to remember. P45 = 65nm, X48 = 90nm, it's much more expensive to produce X48 boards, that's why they cost more I guess..



17137.png



If you look at the 1x HD4870 vs 2x HD4870 results, you see that it got more than 100% increase? Which should be impossible?


But then I looked at the system specifications, and AnandTech are using 1x HD4870 on a nForce680 motherboard (16x PCI-Express1.1 that is) and the Crossfire on a Intel X48 (16x PCI-Express2.0)


So it seems like the nForce680i performance is held back by the bandwidth of 16x PCI-Express1.1 (which equals 8x PCI-Express2.0)


So might it be that the HD4870 actually is held back by the bandwidth here?
It's a bit sad if that's true, because P45 is simply superior to X48, and it's quite sad for all those not upgrading their motherboards every now and then as they will even with 16x PCI-Express1.1 lose much performance..
 
Great insight there, RamGuy. This does add another piece of evidence to the pile. I'm glad I have a X48 board on the way (P5E Deluxe).
 
You have to remember. P45 = 65nm, X48 = 90nm, it's much more expensive to produce X48 boards, that's why they cost more I guess..



17137.png



If you look at the 1x HD4870 vs 2x HD4870 results, you see that it got more than 100% increase? Which should be impossible?


But then I looked at the system specifications, and AnandTech are using 1x HD4870 on a nForce680 motherboard (16x PCI-Express1.1 that is) and the Crossfire on a Intel X48 (16x PCI-Express2.0)


So it seems like the nForce680i performance is held back by the bandwidth of 16x PCI-Express1.1 (which equals 8x PCI-Express2.0)


So might it be that the HD4870 actually is held back by the bandwidth here?
It's a bit sad if that's true, because P45 is simply superior to X48, and it's quite sad for all those not upgrading their motherboards every now and then as they will even with 16x PCI-Express1.1 lose much performance..

This is false. He was using a 790i, the 4870 is just that good. It's not a bandwidth limitation of PCIE 1.1.
 
Back
Top