Slashdot Mirror
Quad PCIe Motherboard
SlipKid writes "PCI Express Graphics cards have allowed for some new and innovative ways to increase rendering horsepower in Desktops and Workstations. Recent introductions of NVIDIA's SLI and ATI's CrossFire technology have enabled dual PCIe Graphics cards in a
load-sharing architecture. Motherboard manufacturers are jumping into the fray now and Gigabyte has released a Quad PCI Express graphics enabled motherboard, capable of running four cards at once. The board is not capable of running Quad SLI, mostly due to lack of NVIDIA driver support currently but it does offer support for eight simultaneous display outputs on four Graphics cards."
Well, $topic pretty much says it all. More PCIe-slots, great, but it'd be nice if there were stuff besides graphics-adapters to push in.
Even if you could do Quad SLI, would it make that much of a difference in performance? At what point would splitting the rendering task be more work than it's worth?
Note I didn't say optimal performance or peak effenciey or any other term to make it seem like more cards would just equal "OMFG MORE FPS, YESSZZZ!". No. With games like BF2 that are starting to require specific actual components of stuff coupled with how much things like DirectX are a huge huge factor in games, you are going to need massive amounts of GPU power to get alot of stuff to run.
I mean (not to plug them or anything) but look at games like Project Offset, which plans for real-time rendering of everything no cutscenes nothing. The processing power of that game is going to be astronomical. I bet it will hit at the least a 2 PCI-E card requirement with at least 1.5 or 2 gigs of ram and 3.0+ GHZ processor, probably 3.4+. And we all remember how systematically intense past games like Farcry were, imagine cranking out a game that's five times as powerful as Farcray or even P:O you're going to require so much raw processing power it's insane.
Which itself is within the true nature of computing, technology evolves, advances, grows faster or more powerful or more advanced. I still think it's sad though, I mean you look at some of the top of the line cards these days required for games, they are insanely priced (200,300 even 400-500 or more). And yes while you can go with something slightly slower and save alot of money, as I originally said I think it will hit the point where they simply will not run without X amount of cards or equipment. Just like I can't run modern games like BF2 or HL2 on my current setup, same thing in a few years for people wanting that hot new title that needs quad cards. The price will be fucking outrageous too. You thought $400 for an Xbox 2 was bad, wait until you need to drop $300 per graphics card, two three or four times plus all the other components just to play games.
Nvidia and ATI are wetting themselves awaiting that day. Why sell them one GPU when Game X they want needs quad cards to even execute.
Aw Frell this
I would love to see a quad-Opteron mobo with four x16 PCIe slots but arranged in a way that traffic is spread across all HT links. So that I could use it to put 4 PCIe SATA cards, and have the highest possible read/write I/O throughput for a Linux software RAID array. Hardware RAID is out of the question, since no constructor offers a way to create arrays of disks across 3 or more cards. An Opteron has 3 HT links, 2 of them could be used as coherent links to other CPU's, and 1 of them could be used as a link to an external PCIe bridge chipset. The solution I would like to see implemented is one where 4 PCIe bridge chipsets would be connected to their own Opteron, via their own HT link. And each PCIe bridge chipset could provide at least one 16x slot.
Some numbers: each of the four x16 PCIe bus would allow for 2500 MT/s * 16 bits / 8 = 5000 MB/s of traffic in each direction. And each of the 4 HT links: 1600 MT/s * 16 bits / 8 = 3200 MB/s. The global amount of I/O would be 3200 MB/s * 4 = 12.8 GB/s in each direction ! (HT links are the bottleneck). To resolve this bottleneck AMD would either need to increase their width from 16x16 to 32x32 bits or need to increase the signal freq from 800 MHz to 1.25 GHz (current limit is 1 GHz for coherent links and 800 MHz for the ones facing outside worlds -- chipsets seem to lag a little bit regarding HT frequency).
But for some reason no constructor has ever designed such a board (Tyan only did it with 2 PCIe chipsets on their S2895 mobo). Why oh why is that the case ?! Seems like nobody understands the true potential of HT. This could provide a low-cost solution to so many perf issues I have seen in the various companies I have worked for... Argh !
I'm not sure you're right.
:-)
How much power do you need for your 8 PC's at home?
Let's assume no more than one of them is actually a gaming rig.
Let's assume 2 more are MPEG-4 decoding boxes.
Let's assume another two run office apps. All concurrently.
Prioritize your processes properly and a dual-core dual-processor rig will do this with modern mid-range processors.
You can even do this with windows using Jetway's Magic-Twin (I do this with 2 seperate consoles and WinXP in my car).
Further, due to all your harddrives being piled in one place serving everyone, you get both a RAID-5 volume, a secondary backup volume to back up your entire RAID5, and if you really want to you can go RAID6 as well. And you get to pool everyone's unused space together, greatly optimizing disk usage.
RAM will be in demand, but not really a problem you can't solve with 4 1-Gig DIMMs stuck in, (and a Gigabyte I-RAM with another 4G for swap if you really want to go overboard... though I'd wait for the SATA2 version).
Another great benefit is QUIET. the machine will be stuck away somewhere and make a lot of noise. Fans, drives, the works. your 8 workstations though will be silent as a grave.
There's several other quirks you'd have to work out such as external peripherals (USB2 hubs wherever applicable), packet shaping for that 15-year-old daughter who wants to run P2P apps, and you'd have to keep the system clean of adware or else.
All in all, for the amount of money 8 new entry-level home PC's would cost, I could build a hydra that would knock the socks of your home box both in data reliability, speed, storage space, noiselessness and bragging rights, whereas availability stretches both ways (lose the mobo and you're fucked all the way, but lose a DIMM, lose a CPU and your box is slightly slower till you get it replaced, lose a drive and you don't even feel it). Performance-wise it'd rock too, as most of the users are not using disk I/O most of the time, and a simple software SATA raid5 (or even a H/W one) with new drives would easily go into the 200-400MB/sec ballpark and when only one of the users is doing something that needs disk I/O it'd fly.
Build it around a 3U or 4U chasis with server H/W and you're set
-
I don't know why, but since i'm on /. and it includes support for more displays i automaticly think of porn. Maybe it's the crowd ....
3DFX SLI = Scan Line Interleaving
Nvidia SLI = Scalable Link Interface
Yes, Nvidia based their version on the ideas they acquired from 3DFX when they bought them out, but the actual techniques they use now are much more advanced. IIRC, the driver does automatic load-balancing, in the sense that if there are more polygons on one section of the screen than another, the rendering will be split so that each card still renders approximately half of them - even if that means one card is doing 75% of the actual screen resolution.
Try 12-14 years ago and SGI instead of 3dfx. SLI is pretty close to the multi-pipe configurations SGI had on their ONYX systems -- generally up to 3 parallel reality engines in a single machine.
Of course, that machine cost upwards of $700k. But multiple CPUs (2,3,4) were pretty typical.