Printable View
Thats what this thread has been about. More than one eye brow has been raised. He was referring to the old chipset. Not to many on this thread are looking at that chipset. Were discussing the Rd600. . He simply pointed out the many faults that chipset had. I am glag they finely got it fixed as you say. But its a little late.Quote:
Originally Posted by Praxis1452
No, that is why they have all that on board RAM to work out of, the PCIe bus is mostly for communicating with the CPU and to "page" data out when its not immediately needed but still has to be available quickly. Its only on the low end cards like the 6200/7300 with Turbo cache that try to use main system RAM as a frame buffer.Quote:
Originally Posted by Turtle 1[/quote
No, we're already CPU limited by most things these days, not PCIe bus limited.Quote:
Originally Posted by Turtle 1
Did NV ever say why?Quote:
Originally Posted by Turtle 1
That'd make no sense, why release a chipset that won't even run their flagship video card in peak benchmarketing mode? Intel wouldn't be locked out of the dual GPU mobo market either BTW (small loss IMO), don't those eVGA i680 mobos have dual x16 PEG's?Quote:
Originally Posted by Turtle 1
This is neither here nor there and speculative to boot.Quote:
Originally Posted by Turtle 1
Also, quote tags FTW!
The odds of R600 using the full bandwith of PCIe 8x are very very low.
No, that is why they have all that on board RAM to work out of, the PCIe bus is mostly for communicating with the CPU and to "page" data out when its not immediately needed but still has to be available quickly. Its only on the low end cards like the 6200/7300 with Turbo cache that try to use main system RAM as a frame buffer.Quote:
Originally Posted by mesyn191
No, we're already CPU limited by most things these days, not PCIe bus limited.
Were not talking about a released GPU are we? So until their tested at 16x vs. 8x we don't know. Thats what the gentleman was asking . About future proof. Thats why I am waiting for Bearlake x38 with DDR3 support.
Did NV ever say why?
Not that I know of ut we can't test on an NV sli 8x PCI-e M/B can we
That'd make no sense, why release a chipset that won't even run their flagship video card in peak benchmarketing mode? Intel wouldn't be locked out of the dual GPU mobo market either BTW (small loss IMO), don't those eVGA i680 mobos have dual x16 PEG's?
I don't know why they would do such a thing. But we have to wait and see what AMD next Chipset is for AMD before we know don't we?
After looking at the NV chipset results don't want one. I will wait for RD600 on retail but not looking like I want them either.
I was referring to Intel chipsets running XF . If they don't run r600's in XF were locked out of the intel chipset with XF.
This is neither here nor there and speculative to boot.
Also, quote tags FTW![/QUOTE]
Yep it is . But who knows I don't niether do you
nay....same problems as before....it is ULI's fault though....not DFI's...just the fact that DFI ignores it bothers me.Quote:
Originally Posted by Praxis1452
Would ATI come out with a video card that requires 2 true 16x slots? Nvidia did that with the 8800 right? I am just asking... not getting smart.Quote:
Originally Posted by mesyn191
And google also would have shown me 1000's of reviews from people that I don't think know what they are talking about....I do view XS.org opinions higher than the rest. Especially from established members.Quote:
Originally Posted by mesyn191
Quote:
Originally Posted by mesyn191
I got worked up because I was asking a serious question and in the amount of time it would take to berate someone I could have recieved an answer. I should probably get out of the Army too! :-)
If we're already CPU limited with what is out right now how will having a more powerful GPU make us less CPU limited and cause us to become PCIe bus limited? Future proof is an impossibility in the PC world BTW, or for that matter for anything, hell you can't even reduce to the point where it becomes unlikely....Quote:
Originally Posted by Turtle 1
Who cares if you can or can't? Without knowing the precise reason why NV chose to do that with the G80 all you can make is half assed assumptions, perhaps there really is something going on to require x16 lanes per PEG or perhaps there is some G80 specific bug that they felt they didn't have time to bother with to fix and so just released as is stating it as a requirement. Just because NV does something does not mean that AMD has to do it too or that it is likely AMD will do it.Quote:
Originally Posted by Turtle 1
Who cares what AMD's next chipset is for AMD, we're talking about Intel, and even if we weren't talking about Intel how would that do anything at all to support your conspiracy theory that they're trying to lock Intel out of the dual GPU market?Quote:
Originally Posted by Turtle 1
But see, no one cares what you do or don't want, the point is its available and it does have dual x16 lane PEG's so Intel would certainly not be locked out of the dual GPU market.Quote:
Originally Posted by Turtle 1
You sure about that? Don't those R600 have the dual dongle bridge that internalizes the whole Crossfire standard a la the new X1950 Pros?Quote:
Originally Posted by Turtle 1
Sound pointless to bring it up then, and pretty far off topic which we've already done enough of so far, so enough of that mmkay? Go start another topic if you want to discuss possible DoJ investigation outcomes.Quote:
Originally Posted by Turtle 1
Well to my knowledge you got 20GB/S on a 8X. I don't know if that is enough though for a R600.
Could they? Sure. Would they? Possibly, but only for Crossfire would this be a requirement.... You going to use Crossfire?Quote:
Originally Posted by 2000army
If you don't know enough about the tech. that you have to ask here to find out about it then how can you evaluate the merit of those reviews that you can find with google? Site layout, grammar? This site is more for showcasing peoples' crazy OC's, bench testing, hardware reviews, and the occassional teaser leaked pics of unreleased hardware a la Conroe and 65nm K8's.Quote:
Originally Posted by 2000army
Alright fair enough. You want me to delete my pertinent comments?Quote:
Originally Posted by 2000army
http://www.microsoft.com/whdc/device..._graphics.mspxQuote:
Originally Posted by Poodle
x8 PCIe is 2Gbps send/receive, on par with AGP 8X apparently.
If crossfire shows a significant increase by using it....then I would like to have the option without having to worry if my mobo meets the requirements....chances are that this is the last ATI/Intel motherboard so I would want it to last.Quote:
Originally Posted by mesyn191
If you want...that is up to you.Quote:
Originally Posted by mesyn191
Thanks,
Eric
....& has the Penny dropped yet??Quote:
Originally Posted by mesyn191
Cos 16X PCIe speed is exactly the same cr4p as 8X AGP over 4x AGP....IT MADE ABSOLUTELY NO DIFFERENCE WHAT-SO-EVER!
Even Tony has said about this if you actually read this thread back a while!
We should stay on topic & wait for more news. Better than getting Modded!
Too many threads RE: this Chipset have been closed for Lame O.T. posting/Spamming/Phreaking/Outrages & personally, I don't want to see this one get terminated either. :stick:
Lets all be like little "Fonzies" & give Tony & Co. chance to get over their sickness in peace. :cool:
can i say this is another delay ? :(
Has anybody heard anything about the physics solution for this board?
I haven't seen too much besides a few press releases from ATI and Havok relating to the concept. I would like to see a test or even better a comparison with Aegia PhysX in the upcoming reviews. I just hope that support doesn't completely evaporate and this ends up becoming one of those features that never comes to fruition.
I'd be very happy shifting my X1900XT to physics if and/or when it becomes available, if it proves worthwhile. :)
Some basic numbers as we finish up the short preview for release later today. In-depth review on Tuesday.
E6600
ATI X1950
2GB OCZ Flex XLC - DDR2-800 3-3-3-9 (1T for RD600/680i)
Raptor 150
SuperPI 32M-
DFI RD600 - 20.42
Intel 975X - 20.01
ASUS 680i - 20.18
3D06 -
DFI RD600 - 6442
Intel 975X - 6416
ASUS 680i - 6423
Quake 4 - 1280x1024 HQ
DFI RD600 - 110.8
Intel 975X - 109.7
ASUS 680i - 110.5
Nice work, but do you also have results for encoding and other apps like that?
nice work Bingo........ it is possible that you make super tests of the devout ones with command rate 3t ? sorry for my english :D
2T - 20.53 3T - 21.13Quote:
Originally Posted by danysm73
We will have a Nero Recode test in our preview today and full audio/video tests in the final article.Quote:
Originally Posted by danysm73
bingo13
Will it be published on anandtech ?
Looking at these figures, you'll have to be impressed with Intel. A two year old 975X chipset is still hanging in there with the "latest and greatest" from the competition. Makes me wanna hug my AW9D-Max (which is only going to get better with the 1333 strap bios).Quote:
Originally Posted by bingo13
Isn't PCIe serial, so duplex bidrectional bandwith.Quote:
Originally Posted by mesyn191
That makes x8 PCIe 4GB/s (duplex) and x8 AGP 2GB/s....Making x16 PCIe 8GB/s (duplex). Although a graphics card mostly reads data, simplex, so a moot point.
http://en.wikipedia.org/wiki/PCI_Express
http://www.pcisig.com/specifications/pciexpress/
And I read somewhere the 8800GTX needed x24 AGP total bandwidth, can find it now so I might be wrong about that. What I do know is one guy I was talking to, had his in the x8 lower slot of an Ultra-D and his 06 score went from 8000 to 11500 moving it to the x16 upper slot.
you have to remember there is data transfer over the bridge or dongle...which basically puts 8 to each card and 8 over the dongle if you wish.
I remember talking about dongle less Xfire a while back and how efficient it was on RD480 and the answer was quite a bit less than RD580 which is 16x16 but identical to RD580 if both chipsets were using a dongle.
I think it boils down to this, you run a dongle and a bridge you will see no loss.
Also, for those who may doubt this i have a question for you. A friend of mine tests many boards and ram for me, he did some tests with DFI 590SLI and the Ultra II M2 which is NF4 but tweaked for Am2 and DDR2, the 8x8 NF4 board was faster in SLI with the same 7900's used on both boards at the same exact clocks...in fact in some instances 10% faster ;) so why is it?
I don't worry about 8 x 8 or 16 x 16 just yet, as long as their is some kind of bridge there is no issue.
I dont think so.. it can be a bug also, because how do you explain that in 3D & max fsb is faster the RD600 and only the Pi not, (yes we dont see brandwith tests, its an other question) but there is the 3T and the asynch memory option too that has got more tweaks that we know this time I think and nothing to say about the other extras what the DFI LP series has only..Quote:
Originally Posted by Zucker2k