Sign in with
Sign up | Sign in
Your question

AMD Crossfire Xpress 3200 Chipset for Core 2 Rolls, But Doesn't Rock

Last response: in Motherboards
Share
February 13, 2007 12:08:21 PM

The new AMD chipset, also known as RD600, keeps pace with Intel's P965/975X and Nvidia's nForce 680i SLI, but is lacking in terms of features. Great overclocking results and a superb motherboard by DFI save the day, fortunately.
February 13, 2007 1:05:30 PM

Smells kinda fishy that no one can make a decent onboard raid controller for intel. Neither Nvidia nor ATI can get the numbers that Intel can with the ICH. Think someone needs to shine a light on this problem!

All-in-all the performance is equal to everything else. Would like to see some of those new physics cards come out so we can make use of that third PCIe graphics slot some time soon.
February 13, 2007 2:24:07 PM

965P comes out on top again.

I think Kudos need to go to Intel for releasing such a kick-a$$ chipset (less PATA support :(  )



On the note of Physics chips, this is something I'd like to see on-board, like LAN or USB, or something that will use an extra vid card.

Honestly, for $250, I can drop that into another vid card and get GFX performance when Physics aren't an issue.

Honestly, I still fail to see the need to seperate Physics hardware, especially with more cores and MUCH more ALU performance coming out of CPUs.

PhysiX came a day late and a dollar short.

I still hope to see AMD come out with something where I might drop in a PhysiX PPU, a CPU, and a GPU, and have them all sit happily on HyperTransport.

Those hopes have dwindled with what I've seen so far.

DAAMIT :evil: 

(lol)
Related resources
February 13, 2007 2:52:56 PM

Hmmm . . . it's not cheaper than current offerings, not faster and has some rather eyebrow-raising I/O bottlenecks. Couple that with (based upon past ATI statements about not being a chipset company) a questionable future of responsive BIOS support and one has to wonder what the draw will be.

I mean, be faster, cheaper, or better supported - pick at least one - and *maybe* there's market opportunity. To be none of the above and to be substantially more expensive than other offerings that ARE fast and well supported (e.g. Gigabyte DS3) just doesn't bode well.

Kudos to DFI for finding creative ways to shine this up beyond the rather steamy pile it appears that ATI gave them.
February 13, 2007 3:47:07 PM

I purchased my new system last week, should get it this thursday, see below.

I first wanted to get that DFI board for it's OCing potential. I guess you all know I'm happy I didn't :D  . All I can say is thanks TOM for that preview a few weeks ago that was spot-on. That's when I decided to rethink my decision and have no regret.

For those interested, stay tune for my OC results next week, altough don't expect result out of the ordinary. I prefer long term over short term, so I'll stay under 0,05V overvolt and less than 55-60C temp to make sure I won't have a dead cpu in 2 years from now, altough I'll probably upgrade before then.
February 13, 2007 4:17:13 PM

Looks good. I would have liked to seen if the extra cost of the Promise SATA controller were worth the extra cost by raid benchmarks. This board gets beat by the 965 but holds its own or even beats its only crossfire rival the 975. Better OCing than 975 for crossfire sounds like a solid market with the R600 comming. By march motherboards with this chipset should be less than $150 so its going to take market from both the 975 and nvidia. Looks to me like AMD/ATI plans to take the the crossfire chipset market leaving Intel with only 965 market. Intel want let this stand so maybe will get a better OCing 975, RAID5, crossfire in the future.
February 13, 2007 4:47:37 PM

And when did this motherboard come out???

Feels sort of late...

Also find it weird that Intel's BadAxe motherboard has 3 PCI Express for graphics... wonder if you can run crossfire and physX with that?

*sighs* If it can, then DFI shot themselves in the foot with this seeing as the chipset can barely even keep up with the old 975 chipset.

(my opinion) :oops: 
February 13, 2007 10:08:08 PM

How funny that so many people, maybe ATi enthusiasts or maybe not, we're hailing this motherboard as the second coming almost promising massive overclocks and better performance than any other motherboard at the time. well to all those people this must be a huge slap in the face cause unless there is some other "RD600" thats coming out this one fails on almost every account .

There really isn't much reason to go with this board until ATi comes out with its next generation of graphic cards, and even then the lack of dual x16 PCIe(whether its used or not) and poor raid performance put this one in the back of the pack

oh and intel's new chipsets that also support crossfire will most likely oc like the p965 or better and bring more support for future cpu's will be coming out soon enough to make this board obsolete
February 13, 2007 11:54:01 PM

This article reaffirmed my plans to buy a 965 board...
February 14, 2007 12:25:30 AM

DFI could have added alot more to that board.
February 14, 2007 3:24:33 AM

Quote:
DFI could have added alot more to that board.


I beg to differ. This board is an absolute dream. I am not sure what THG's problem was with the FSB, but anandtech.com and other reputable people achieved the maximum FSB allowed in the BIOS, 511 FSB. That puts the FSB quad pumped at 2044... right up there with 680i... if not besting it. Also, the DFI BIOS is a god send to uber tweakers. This board is not for the faint of heart, it has more settings to screw up than any other board I have ever seen.

Granted no RAID 5.... but why use RAID 5 on a home board integrated controller.... its a valid argument but sort of nit picking. Also the independent memory clocking is beautifully done. Purely independent. No more compromising between FSB and memory clocks.... clock both to the max at will. Built in MemTest is nice. It recovers from failed OC's perfectly.

As for the USB performance.... if that is what you purchase a mobo for... then I am sorry you have issues. Honestly, I don't have a single thing plugged in to my USB slots other than mice and keyboard and the occasional ipod alternative, so in reality I don't care about USB 2 perf as long as it works reasonably well. I know it sounds like a defense response, but so many over look it and call the 965 the pure champ, when this board can do just as well, if not better in some areas.

965 is solid... but so is DFI, they don't kid around.

Also, anandtech had a review on this board out in mid december.... what took THG so long? This board has been available for about 6 weeks. I realize it is way overdue... but it seems to have been done correctly.

Lastly, the FSB limitation.... is BIOS IMPOSED not chipset limited. :wink:

Side note: Intel does have the best RAID controller.... *sigh* can't win them all lol. We concede to Intel and others on this one.... crappy I/O. :( 

Another review....
February 14, 2007 3:22:41 PM

Thanks for the link :!:
February 15, 2007 1:11:16 AM

Quote:
Thanks for the link :!:


You're welcome. I do my research before I dump $250+ into a mobo :wink:
February 15, 2007 4:03:15 AM

Quote:
DFI could have added alot more to that board.


I beg to differ. This board is an absolute dream. I am not sure what THG's problem was with the FSB, but anandtech.com and other reputable people achieved the maximum FSB allowed in the BIOS, 511 FSB. That puts the FSB quad pumped at 2044... right up there with 680i... if not besting it. Also, the DFI BIOS is a god send to uber tweakers. This board is not for the faint of heart, it has more settings to screw up than any other board I have ever seen.

Granted no RAID 5.... but why use RAID 5 on a home board integrated controller.... its a valid argument but sort of nit picking. Also the independent memory clocking is beautifully done. Purely independent. No more compromising between FSB and memory clocks.... clock both to the max at will. Built in MemTest is nice. It recovers from failed OC's perfectly.

As for the USB performance.... if that is what you purchase a mobo for... then I am sorry you have issues. Honestly, I don't have a single thing plugged in to my USB slots other than mice and keyboard and the occasional ipod alternative, so in reality I don't care about USB 2 perf as long as it works reasonably well. I know it sounds like a defense response, but so many over look it and call the 965 the pure champ, when this board can do just as well, if not better in some areas.

965 is solid... but so is DFI, they don't kid around.

Also, anandtech had a review on this board out in mid december.... what took THG so long? This board has been available for about 6 weeks. I realize it is way overdue... but it seems to have been done correctly.

Lastly, the FSB limitation.... is BIOS IMPOSED not chipset limited. :wink:

Side note: Intel does have the best RAID controller.... *sigh* can't win them all lol. We concede to Intel and others on this one.... crappy I/O. :( 

Another review....

The reason for the VASTLY increased interest in RAID (especially SATA RAID) is the availability of big cheap SATA hard drives (such as the Hitachi DeskStar and Seagate's *smaller* Barracudas). Four such 500 GB drives and you have a 2 GB array. The big problem is that everybody (including Promise) has to compete with Intel's killer SATA RAID solution (which, oddly enough, has its roots in the long-dead ICH5R SATA/RAID solution of S478 Northwood yore).

You don't have a printer? The biggest use for USB (other than keyboards, mice, iPods/MP3 players, and webcams) is for printers; in fact, most folks have *at least five* USB devices. I use a PS/2 mouse and keyboard and have *four* USB 2.0 ports; all four are in use, too (RF remote, webcam, MP3 player, printer). Unless you go with some sort of multifuction device (which I still refuse to recommend for anyone for any reason), you will want extra ports for things like scanners and printers (in my case, more so printers than scanners). Because of the increased need for more USB ports, the *ten* that the DFI supports (tied with the nForce 680i for the most USB 2.0 ports in the Intel chipset camp) is definitely a good thing.

While the RAID support trails Intel (but then, so does everyone else) the support for SATA devices other than RAID is second to nobody (including Intel). The big problem is that SATA is, at best, a point-to-point connection strategy (one device per port). Given two optical drives and a RAID array of four drives, you need six ports. And in the case of most 965-based motherboards, you *can't* bring any PATA devices with you, because they aren't supported; worse, most 975X-based mobos won't support that many SATA devices. Heaven help you if you dual-boot and want to have each OS on its own drive and separate from the pair of OS-specific two-drive arrays (this is proving to be a wanted option if you dual-boot XP and Vista). That means *three hard drives* per operating system. If you go all-SATA (which is now possible) that is *eight* SATA ports. No Intel chipset supports that many SATA ports without help (the same is equally true of nForce 680i). That's where the motherbaord manufacturers separate themselves from each other.

THG's quibbling isn't with DFI. The problem THG has is more that the chipset is facing uncertain support from a company (AMD) that has expressed a determination *not* to remain in the chipset business, especially supplying chipsets for its competition.
February 15, 2007 5:52:28 AM

Quote:
I beg to differ. This board is an absolute dream. I am not sure what THG's problem was with the FSB, but anandtech.com and other reputable people achieved the maximum FSB allowed in the BIOS, 511 FSB. That puts the FSB quad pumped at 2044... right up there with 680i... if not besting it.

except that Tony from OCZ says that there is no point running at that as the memory performance & hence overall performance starts dropping from ~470 & advocates a CPU with a multi that lets you max core clock whilst staying under that.

Quote:
The problem THG has is more that the chipset is facing uncertain support from a company (AMD) that has expressed a determination *not* to remain in the chipset business, especially supplying chipsets for its competition.

afaik AMD does intend to stay in the chipset business but yes, not for Intel CPUs.
February 15, 2007 10:01:25 AM

Anyone know how well it performs with the QX6700?
February 15, 2007 11:25:57 AM

Quote:
Hmmm . . . it's not cheaper than current offerings, not faster and has some rather eyebrow-raising I/O bottlenecks. Couple that with (based upon past ATI statements about not being a chipset company) a questionable future of responsive BIOS support and one has to wonder what the draw will be.

I mean, be faster, cheaper, or better supported - pick at least one - and *maybe* there's market opportunity. To be none of the above and to be substantially more expensive than other offerings that ARE fast and well supported (e.g. Gigabyte DS3) just doesn't bode well.

Kudos to DFI for finding creative ways to shine this up beyond the rather steamy pile it appears that ATI gave them.


This pretty much sums it up. For the money they are asking for other vendors do better - a lot better. Most 975 Boards are cheaper and offer the same things. PCIe 2.0 is on its way, but instead of keeping up with Nvidias 2x 16 PCIe Links, AMD takes a step back and splits a single x16. If they issued a statement saying that x8 will be enough for any dx10 crossfire card they´ll manufacture, i wouldn´t argue, but i have my doubts. :?
I hope there is a follow up chipset coming soon. This one is probably just for their platform business.
February 15, 2007 11:52:32 AM

Quote:
The reason for the VASTLY increased interest in RAID (especially SATA RAID) is the availability of big cheap SATA hard drives (such as the Hitachi DeskStar and Seagate's *smaller* Barracudas). Four such 500 GB drives and you have a 2 GB array. The big problem is that everybody (including Promise) has to compete with Intel's killer SATA RAID solution (which, oddly enough, has its roots in the long-dead ICH5R SATA/RAID solution of S478 Northwood yore).


I realize that hard drive prices have come way down and sizes are way up, that is obvious. However, Just because drives are cheap doesn't mean that RAID is a good idea. RAID is one of the most oversold functions on mobo's these days and it offers little performance benefit to the *average* end user. Oh, and there is RAID 5, it is just provided by the Promise controller, not the ATI SB.

Quote:
You don't have a printer? The biggest use for USB (other than keyboards, mice, iPods/MP3 players, and webcams) is for printers; in fact, most folks have *at least five* USB devices. I use a PS/2 mouse and keyboard and have *four* USB 2.0 ports; all four are in use, too (RF remote, webcam, MP3 player, printer). Unless you go with some sort of multifuction device (which I still refuse to recommend for anyone for any reason), you will want extra ports for things like scanners and printers (in my case, more so printers than scanners). Because of the increased need for more USB ports, the *ten* that the DFI supports (tied with the nForce 680i for the most USB 2.0 ports in the Intel chipset camp) is definitely a good thing.


Nope, no printer connected. I have no use for it. Rarely do I ever have to print anything, and when I do, I use my networked printer that has a built in Ethernet jack. You also aren't addressing my point. You are addressing demand for number of ports, not the performance of those ports. The only device you listed that would really even care about USB 2 performance is a mp3 player. The other devices wouldn't ever be bottlenecked by 20MB/s performance. The demand for ports is there, but USB 2 performance shouldn't be a critical factor in your purchase choice, especially with the advent of e-SATA for use with external hard drives.


Quote:
While the RAID support trails Intel (but then, so does everyone else) the support for SATA devices other than RAID is second to nobody (including Intel). The big problem is that SATA is, at best, a point-to-point connection strategy (one device per port). Given two optical drives and a RAID array of four drives, you need six ports. And in the case of most 965-based motherboards, you *can't* bring any PATA devices with you, because they aren't supported; worse, most 975X-based mobos won't support that many SATA devices. Heaven help you if you dual-boot and want to have each OS on its own drive and separate from the pair of OS-specific two-drive arrays (this is proving to be a wanted option if you dual-boot XP and Vista). That means *three hard drives* per operating system. If you go all-SATA (which is now possible) that is *eight* SATA ports. No Intel chipset supports that many SATA ports without help (the same is equally true of nForce 680i). That's where the motherbaord manufacturers separate themselves from each other.


Ok. I agree with that, but how many people use all 8 ports? Yes, it is nice to have and the DFI board does have them, but how many use them. Also, most people still use PATA optical drives, not many use SATA. The use of SATA optical drives is on the rise, as is the addition of PCI express lanes added to support the increased bandwidth. SATA was designed as a point to point connection for a reason, so two devices don't have to share bus bandwidth. Now it isn't as much of a problem, but as time progresses it will be come necessary to have dedicated bandwidth. Granted, now there are no devices to max out the SATA 150 spec, but with the bandwidth there, someone is bound to use it. :wink:

Quote:
THG's quibbling isn't with DFI. The problem THG has is more that the chipset is facing uncertain support from a company (AMD) that has expressed a determination *not* to remain in the chipset business, especially supplying chipsets for its competition.


You have to rely on DFI. When they make a board, they go all out. They will keep the BIOS updated and the drivers as fresh as they can. I don't see AMD dropping the ball on this just to piss off Intel, especially when Intel's chips are doing so well. If I were AMD, I would continue to make chipsets for Intel because there will always be a market for them and you can grab up some revenue by making good chipsets for Intel or give up and let Intel/Nv have the entire pie. DFI has a dedicated forum where multiple BIOS revisions have come out of due to end users posts. DFI actually monitors their boards and has official rep's that contact engineers to resolve BIOS level issues and compatibilities. That is where the version of my BIOS that I use on my RD580 board came from, users requests to fix compatibility with certain memory modules.

Quote:
Hmmm . . . it's not cheaper than current offerings, not faster and has some rather eyebrow-raising I/O bottlenecks. Couple that with (based upon past ATI statements about not being a chipset company) a questionable future of responsive BIOS support and one has to wonder what the draw will be.

I mean, be faster, cheaper, or better supported - pick at least one - and *maybe* there's market opportunity. To be none of the above and to be substantially more expensive than other offerings that ARE fast and well supported (e.g. Gigabyte DS3) just doesn't bode well.

Kudos to DFI for finding creative ways to shine this up beyond the rather steamy pile it appears that ATI gave them.


This pretty much sums it up. For the money they are asking for other vendors do better - a lot better. Most 975 Boards are cheaper and offer the same things. PCIe 2.0 is on its way, but instead of keeping up with Nvidias 2x 16 PCIe Links, AMD takes a step back and splits a single x16. If they issued a statement saying that x8 will be enough for any dx10 crossfire card they´ll manufacture, i wouldn´t argue, but i have my doubts. Confused
I hope there is a follow up chipset coming soon. This one is probably just for their platform business.

DX9 in crossfire could have worked on 2 x4 PCI-e lanes. THG did a test and current GPU's only needed about 4x PCI-e bandwidth to perform at their best. While, DX10 are more powerful cards (obviously), do you really think they will use more than 2x the bandwidth of DX9 cards? I don't have the answer to that, but just think it through. If it were true, they would be forcing people into new mobo's and would require them to pick specific ones which sounds a bit ridiculous, but all be it possible. Also, this chipset has been in the works for 2 years (or more I forget), so its roots are from a time when a split x16 lane was normal.

I wouldn't expect a follow up chipset anytime soon. Given AMD's narrow view to cut off Intel support at the expense of profit, this will most likely be their last Intel chipset for a good while. Only time will tell.

Quote:
[quote="SuperFly03]I beg to differ. This board is an absolute dream. I am not sure what THG's problem was with the FSB, but anandtech.com and other reputable people achieved the maximum FSB allowed in the BIOS, 511 FSB. That puts the FSB quad pumped at 2044... right up there with 680i... if not besting it.

except that Tony from OCZ says that there is no point running at that as the memory performance & hence overall performance starts dropping from ~470 & advocates a CPU with a multi that lets you max core clock whilst staying under that.


Could you provide a link please? I am not sure I really believe him, unless a new strap comes into play around 470. It is possible, but I don't know. So please show me some evidence. :D 
February 15, 2007 2:18:52 PM

Quote:



For the money they are asking for other vendors do better - a lot better. Most 975 Boards are cheaper and offer the same things. PCIe 2.0 is on its way, but instead of keeping up with Nvidias 2x 16 PCIe Links, AMD takes a step back and splits a single x16. If they issued a statement saying that x8 will be enough for any dx10 crossfire card they´ll manufacture, i wouldn´t argue, but i have my doubts. Confused
I hope there is a follow up chipset coming soon. This one is probably just for their platform business.


DX9 in crossfire could have worked on 2 x4 PCI-e lanes. THG did a test and current GPU's only needed about 4x PCI-e bandwidth to perform at their best. While, DX10 are more powerful cards (obviously), do you really think they will use more than 2x the bandwidth of DX9 cards? I don't have the answer to that, but just think it through.

The question isn´t really if x8 is enough. Would you run a 8800GTX on a PCIe x8 slot? It´s a high end card and x8 doesn´t sound like high end at all. The new Radeon is rumored to have even more memory bandwidth than the G80. Right now it would probably work, but in 6 months? How about in 12 months? Someone that buys 2 Crossfire cards is what i´d call an enthusiast and that bunch of people tend to go after the greatest and latest (and fastest!). So the board will probably see two or three generations of GPUs before it will be phased out. Two x8 slots isn´t great, it isn´t even fast - it´s just late.

Quote:

If it were true, they would be forcing people into new mobo's and would require them to pick specific ones which sounds a bit ridiculous, but all be it possible. Also, this chipset has been in the works for 2 years (or more I forget), so its roots are from a time when a split x16 lane was normal.

Yep, it´s from a time where horses still were the fastest way to travel and people thought the world was flat. Its old. Intels 975 chipset offers the same features (except the excessive overclocking) and are cheaper and more mature. I´m not saying the RD600 is a bad product. I´m just saying i´m not seeing it´s market.

Quote:

I wouldn't expect a follow up chipset anytime soon. Given AMD's narrow view to cut off Intel support at the expense of profit, this will most likely be their last Intel chipset for a good while. Only time will tell.

I hope they do a follow up, but i´m with you on that - i doubt they will.
February 15, 2007 8:13:48 PM

people keep saying you need 1080p even though you don't notice the difference between 1080i/p on good tvs but people will still complain about a tv that doesn't have 1080p capability...
February 15, 2007 10:10:40 PM

I haven't seen nor heard of any tests that have measured how much an 8800GTX saturates the 8x speed or 16x and if there were any performance hit when running sli on 16x vs 8x but it would be interesting to see. i mean even though we not surpassing 8x yet with the further development of graphic cards, games and dx10 you wouldn't want to be the one that didn't get the motherboard that only supported 8x then take a performance hit only because you got what was satisfactory at that time, i mean everyone wants to make there computer future proof in some way or another and when other high end motherboards support 16x in dual card mode why shouldn't this one...

i know i am still keeping in mind this has no effect right now and my scenario is far from what could actually happen, but you know its a high end $250 mobo it should have everything i want/need and then some right?!
February 16, 2007 12:14:34 AM

Quote:
ok slobogob and you need to please please understand something. you quote that article and yet miss the point.

gfx cards are output devices, they have nothing, i.e. zero to do with PCI-e utilisation unless someone can prove me wrong.


That isn't true in the slightest.



This is the 975x chipset. Look at the first link, it is the FSB. This is the communication between the CPU and the northbridge. Attached to it is the x16 PCI-e graphics card lane. Data is passed to and from the GPU from the CPU and RAM (depending on data load and location). From the northbridge the data is transmitted to the GPU via a x16 link (as I have said), thus the more data that is shunted into the GPU the more the x16 lane is utilized. So in the case of 8800GTX, as games move to DX10 and shunt more instructions to the GPU through the NB, the more bandwidth is required. From there there GPU does its processing and outputs the image via DVI (in most cases these days). That is how it utilizes the PCI-e lane.

Quote:
how powerful a gfx card is means it will render more of the total amount of frames sent to it by the CPU until it becomes CPU limited.


So how does that data get there? :wink:

Quote:
i do not know how someone managed to perpetuate the myth, especially on a tech forum that more powerful gfx cards need more bandwidth.


The more powerful the GPU, the more quickly it can process information. The faster it processes information the quicker it will need new information. The quicker it needs new information the quicker the CPU will need to transfer that information, and it gets there via the FSB and x16 link (in the 975x case).

Quote:
now, dx10 games may need more bandwidth to render all the frames and indeed if there is any physics or other data that needs to be communicated back to the CPU that might mean there will be a need for the bandwidth 8x or 16x offers but in no way will the ability of a gfx card to render frames contribute to the amount of data being sent to them.


It isn't just the amount of data, its the data rate that matters. 3GB of data is 3GB of data, but do you transfer it at 10Mb/s or 1000 MB/s, there isa vast difference.

Furthermore, if you go bridgeless on SLI, the data gets transferred between the cards via the PCI-e lanes as well, further increasing the bandwidth need.

Maximum PC did a benchmark (as did other PC mags) and found that X8 SLI vs X16 SLI on each lane had no performance benefit or hindrance (with in the given margin of error).

Quote:
i know i am still keeping in mind this has no effect right now and my scenario is far from what could actually happen, but you know its a high end $250 mobo it should have everything i want/need and then some right?!


It depends more on what you need than if it has everything made. You do research to develop expectations and the ability to target your needs. A $250 should have everything you need, you are right, but what do you need? That is a question only you can answer.

Quote:
people keep saying you need 1080p even though you don't notice the difference between 1080i/p on good tvs but people will still complain about a tv that doesn't have 1080p capability...


This is subjective, like Stranger said. If you buy a new TV, definitely go 1080p, but if you already have 1080i, you most likely won't notice it (depending on how far you are from the TV :wink:) .
February 16, 2007 12:24:34 AM

Quote:



For the money they are asking for other vendors do better - a lot better. Most 975 Boards are cheaper and offer the same things. PCIe 2.0 is on its way, but instead of keeping up with Nvidias 2x 16 PCIe Links, AMD takes a step back and splits a single x16. If they issued a statement saying that x8 will be enough for any dx10 crossfire card they´ll manufacture, i wouldn´t argue, but i have my doubts. Confused
I hope there is a follow up chipset coming soon. This one is probably just for their platform business.


DX9 in crossfire could have worked on 2 x4 PCI-e lanes. THG did a test and current GPU's only needed about 4x PCI-e bandwidth to perform at their best. While, DX10 are more powerful cards (obviously), do you really think they will use more than 2x the bandwidth of DX9 cards? I don't have the answer to that, but just think it through.

The question isn´t really if x8 is enough. Would you run a 8800GTX on a PCIe x8 slot? It´s a high end card and x8 doesn´t sound like high end at all. The new Radeon is rumored to have even more memory bandwidth than the G80. Right now it would probably work, but in 6 months? How about in 12 months? Someone that buys 2 Crossfire cards is what i´d call an enthusiast and that bunch of people tend to go after the greatest and latest (and fastest!). So the board will probably see two or three generations of GPUs before it will be phased out. Two x8 slots isn´t great, it isn´t even fast - it´s just late.

I agree it doesn't sound fast, however the question is does it work and can it survive the life of the build. If the GPU's outpace x8 in 2 years, I would be surprised, but it is possible. Your arguement is generally sound, but it depends on future tech, and we can both only guess about it at this point. I do see your point though. :) 

Quote:

If it were true, they would be forcing people into new mobo's and would require them to pick specific ones which sounds a bit ridiculous, but all be it possible. Also, this chipset has been in the works for 2 years (or more I forget), so its roots are from a time when a split x16 lane was normal.

Yep, it´s from a time where horses still were the fastest way to travel and people thought the world was flat. Its old. Intels 975 chipset offers the same features (except the excessive overclocking) and are cheaper and more mature. I´m not saying the RD600 is a bad product. I´m just saying i´m not seeing it´s market.

DFI has always had a small market. Their motherboards are more about targeting a small overly aggressive market. I agree it isn't a big market, but it is a highly focused target market.

Quote:

I wouldn't expect a follow up chipset anytime soon. Given AMD's narrow view to cut off Intel support at the expense of profit, this will most likely be their last Intel chipset for a good while. Only time will tell.

I hope they do a follow up, but i´m with you on that - i doubt they will.

Only time will tell. :) 
February 16, 2007 6:45:47 PM

Quote:
i like the fact that for once someone has tried to back up what they say however, i do not agree with it yet. how much info is sent it limited by the CPU.


In general that is true. It really depends on the setup. Some systems are limited by CPUs others are limited by the GPU. You'd have to test each build and get down and dirty to figure out which it is for each system.


Quote:
can you show me a link as to where you got your info from.


It is mainly from reading up on general chipset info, however the GPU bottleneck can be found here

Look at the first graph, you will note that the 8800GTX's FPS grew drastically on the lower res settings when the platform was switched from AMD to Intel. The result was astonishing. Now, once the resolution is increased and more load and processing is required by the GPU, the system becomes GPU bottlenecked as opposed to CPU bottlenecked. So, even within the same system it depends on the resolution and quality settings whether it is GPU or CPU bottlenecked.

Quote:
not that i mean to offend you but could you provide a link please.


No offense taken at all. I don't mind my views being challenged, as long as the discussion is kept respectful. :D 

Quote:
oh and again thanks for being the first person who has tried to back up there view in response to one of my posts. so many just don't respond and i will be glad to admit i am wrong if that is the case.


I love a good discussion and defending my views, because I can be wrong as well and I always learn something new.
February 16, 2007 9:57:36 PM

Quote:
Look at the first graph, you will note that the 8800GTX's FPS grew drastically on the lower res settings when the platform was switched from AMD to Intel. The result was astonishing. Now, once the resolution is increased and more load and processing is required by the GPU, the system becomes GPU bottlenecked as opposed to CPU bottlenecked. So, even within the same system it depends on the resolution and quality settings whether it is GPU or CPU bottlenecked


sorry, but that is not quite what i was looking for.

of course the low res FPS will increase with the AMD to intel switch. this has nothing to do with the gfx card only the CPU's ability to produce at higher amount of frame data to be sent to the gpu.

Exactly. The GPU is what is held constant, so the ability of the computer to draw frames is the same on each system, right? If that is so, then the increase in FPS can not be attributed to the GPU's ability to draw data, but rather the increase is attributable to the increase in CPU power. With the increase in CPU power, it is able to issue instructions to the GPU more frequently and from the the GPU can derive frame information and draw more frames.

Quote:
at low resolution gfx cards can draw them more quickly and so produce more frames.


Agreed, to the point where the CPU isn't sending enough instruction for the GPU to process.

Quote:
the lower you go the easier it is for them until they are drawing as much as the CPU can process. this does not show whether the GPU is asking for more or just completing more of what is being sent.


Ok, I see your argument. You are saying that the CPU issues a finite set of instructions. If the computer is GPU limited, then the GPU doesn't complete all the instructions, rather just as many as it can. Otherwise, the GPU just processes the full amount of instructions sent by the CPU. Am I understanding what you are trying to get across?

Quote:
i am not doing well putting what i want to say into words but i will need more links because your logic seems wrong. i will try to look up my own links but the fact that pci-e, 8x AGP and 4x AGP all produice the same amount of FPS in games suggests to me at least that either the gfx cards are not asking for more data than what the interconnects can supply or that the amount of data that the CPU provides is the same regardless of the gfx card.


I will go back and find another THG article that has led me to my point of view. If you find any links, please post them and I will give them a read. I see your point. Let us research it!!!! :D 

Quote:
somehow that last statement itself seems to have no logic and probably is complete crap but oh well, will post it anyway.


Many things about computers are hard to put into words.... everyone has issues with it, myself included. :wink:

Edit: Finally found what I was originally looking for.

here
February 17, 2007 3:17:07 PM

Quote:
i might see if suggesting THG do a repeat of this article on modern hardware and do a more detailed analysis of why the results are so. i feel this may be one of the only ways to solve this is to get some experts to look at it.


I agree, I think that is what it will take too.

I will think about the results on the link and my understanding of them and get back to you. Hmmmm such an interesting puzzle.

Edit: I have an idea. What if I did some benchmarking on my system holding everything constant but the CPU speed and tested the FPS of certain games and 3DMark on my system... let me think about what that would give us...
February 22, 2007 6:25:41 PM

Well Im a gamer and I love this board. It may not be the fastest or rape as much face as one would want but Id buy it again just for the indapendent ram clocking and over all stability under a oc. I have been able to obtain 511x8 8 hour orthos stable with my e6600. I have not had a single problem using any feature offered on the board, I wish my old x850 wasnt agp so I could try out the crossfire+phys card. I hope the dfi support hold true with bios revisions and tweaks. I cant stress how much I love setting my ram at 503fsb@4's 12 20 then play with the oc on my cpu and not have to give the ram a second thought.
!