Slashdot Mirror

If this thing really does need 300 watt just to power the card i predict it will go down in flames. The cards by Nvidia and AMD have made major strides on heat/power and Intel wants us to go back to 300 watt GPUs? WTF? I for one don't want or need another Intel space heater, thank you very much.

Not to mention this thing looks different enough from ATI/Nvidia cards that if it gains a foothold we could end up just like the bad old days of Win9x. In case any youngsters don't remember there was this little thing called Glide which was a proprietary 3DFX tech that would only work on their cards, so you ended up with games that looked like shit if you didn't have a Voodoo card or games that supported the other cards but ran like shit on Voodoo. I personally like the fact that I can buy either Nvidia or ATI and have games both past and present work and have comparable graphics. This thing looks to be different enough that I'm betting code not written specifically for it will probably take a serious performance penalty. No thanks.

While I had hopes when it was first announced that Larrabee would be simply a new low power GPU by Intel that wouldn't suck, instead it looks to be another Netburst space heater. Meanwhile with the Ion and the 780 Nvidia and AMD have made great strides in the IGP space that currently is ruled by Intel. Between this and the way Intel seems to be burning their bridges with lawsuits from Nvidia and AMD one really has to wonder just what the hell they are thinking. Intel with their crazy maneuvers has made up my mind and for the first time since the old K2 I'm building an AMD box next month to replace my aging P4. Because with all this craziness who knows if an Intel box would support the GPUs that I want to run? Because just from what I have seen and read so far Larrabee simply isn't the GPU for me. So sorry Intel, but the insanity you have been pulling for the past few months have cost you this life long Intel man. Stick to the CPU and let companies with much more experience with graphics stick to the GPU.

Yes. Toshiba made a Cell-derived coprocessor for those kinds of tasks, and several manufacturers started producing add-on cards for the PC using the chip.

But I thought most of the scientific community and the FLOSS guys were ANTI DRM? Isn't it true you have NO access to the GPU at all and only a limited access to the cell to stop "teh evil piratez!" and enforce Sony DRM? And if the Cell is really that wonderful why wouldn't you simply add a few cells to your x86/CUDA cluster and have a supercomputer capable of doing it all? And since the above cell fits into a 1x or 4x PCIe slot, you could simply add more as required or budgeting allows. So the only advantage I can see for Cell is for making a super cheapo cluster.

But if you are pushing the kind of data that needs a super computer anyway I'm sure there are grants and endowments that will help you get the hardware required for your research if you don't already have the money. And with the above solution you have 1)complete control of the hardware 2)The ability to scale the existing units by adding more CPUs/cells/GPUs, and finally 3) The ability to do several kinds of research as your cluster wouldn't need to be tied to the subset of calculations that the cell outperforms on. Oh and lets us not forget gigabit ethernet to tie it all together to help keep the cluster well fed.

So am I missing something here? Or is all this cell talk just because you can pick them up for cheap at Wally World? And since Sony loses serious money on PS3 sales wouldn't a big uptick in those using the PS3 for research seriously damage the company, who last I heard is already bleeding pretty bad?

"When using the Fast Mode, ASRock advises you not to switch off the AC power to the PC at the mains .. Instant Boot will also only work on Windows systems (XP or Vista) with a single-user account and no password protection"

>The only reason the Wii controller is accurate is that it constantly recalibrates itself from the bar under the display.

Actually no. It uses the bar to track its rotation on its Z axis. Without the bar it can only detect rotation on the Y an X axis. You can test it by connecting it to a pc(you will need bluetooth). http://www.custompc.co.uk/howtos/601059/control-your-pc-with-a-wiimote.html#

;)

A little context might help. This isn't the Inquirer for god's sake.

Gimme and "R"!
Gimme and "T"!
Gimme and "F"!
Gimme and "A"!

What's that spell????:

http://www.custompc.co.uk/news/602910/rumour-control-larrabee-based-on-32-original-pentium-cores.html

This decent round-up here finished off the Raptor line for me. A SpinPoint is much cheaper, much larger and has a lot less noise too. So why bother? Maybe the VelociRaptor will perform better, but I've seen a lot of PC builders get obsessed with the 'badge', i.e. someone told them that their games will run faster with a Raptor...

My favorite print magazine Custom PC (recommended for those PC hardware fetishes, although not sure if they cover the USA?) had a round up of these solid-state devices in the summer:

http://www.custompc.co.uk/labs/601070/solid-state-drives/products.html

Still looks like the market needs more baking by the 1.0 cash-gullible crowd...

What's a little fan noise and power consumption for a few fans compared to a triple cascaded refrigerated rig like this overclock demo.

http://www.custompc.co.uk/news/601310/idf-556ghz-penryn-breaks-three-benchmark-records-in-two-minutes.html

That is 5.56 Ghz demonstrated at the IDF. It ran and broke 3 CPU speed records in under 2 minutes. This is extreme! Don't ask how much power the cascaded refrigeration system uses. ;-)

"Worth then proceeded to show off the fruit of his labours by claiming he could break three world benchmark records in just two minutes. This included SuperPi 1M, where he beat Team Japan's previous record, and he then went on to run AquaMark where his score of 273,000 trashed the previous record of 267,000. Finally, he then ran the 32-bit version of CineBench 10, where a score just shy of 20,000 (he didn't reveal the actual score, unfortunately) again clinched a world record for a quad-core CPU."

Sounds very reasonable to me. TCP/IP is meant to deal with a couple dropped packets here and there, but audio protocols are generally very sensitive.

I wonder how it affects systems where the networking is not on PCI (maybe an integrated northbridge component which is not glued to an internal PCI bus), or the audio controller is on a completely different PCI host controller (this scenario is practically unheard of on most x86 systems though.. would be intriguing to find out nonetheless :)

Ooh. Could it be that these systems are PCI Express and Vista UAA has been coddled to make PCI Express audio not such a bitch? http://www.guru3d.com/newsitem.php?id=3005 although as of last month or so, they seem to have decided they CAN do it without a bunch of the features; http://www.custompc.co.uk/news/115666/creative-unl eashes-pci-express-xfi.html

Here's how some of the specs and prices (Intels from Pricewatch) compare:

E6750:
2.66Ghz
1.30V 4MB L2
1.33Ghz FSB
~$225


E6850:
3.00Ghz
1.35V 4MB L2
1.33Ghz FSB
~$300


6400+:
3.20Ghz
1.35V 2MB L2
2.00Ghz HT
~$240


From what I've read, the E6750 actually outperforms the E6850 since it ran cooler and with less power. So it doesn't look like AMD has a whole lot to offer given the price. Not to mention it doesn't come with a heatsink and fan, something you'd probably have to dish out another $20-50 for.