Intel in the GHz Game Again - Skulltrail Hits 5 GHz

An anonymous reader writes "Intel's Skulltrail dual-socket enthusiast platform has been making the rounds on the web for half a year or so, but we haven't seen many details yet. TG Daily got a close look at an almost complete prototype, which surely sounds almost like a production ready version, judging from the article. Everything that TG Daily describes sounds like Skulltrail PCs will be very limited in availability and insanely expensive. Intel also has said it has developed 'special' Xeon processors with desktop processor attributes just for Skulltrail. These chips are currently running at a stable 5 GHz."

Comments

This looks bad next to a amd dual quad-core system

[Joe The Dragon]

single-slot graphics cards, FBDIMMS and you will need 4 of them to get the max system out of the memory system, SLI useing Nvidia nForce 100 chips over a pci-e x16 1.1 bus split to 2 x16 slots, dual eps power in, 3 chip sets chips that driver up cost and power use.

The dual amd system that this will be like this will use DESKTOP RAM, have 2 or more chipset choices. Also the amd setup lets you have 2 full Northbridge chipsets for even more i/o the nForce 680a uses this and nvidia will likey have a new chipset with pci-e 2.0. The old has a x16 x8 x8 x16 pci-e with a total of 56 PCI-E lanes.

The new amd chipet is also comeing and you may even see a board with 2 Northbridges = 82 pci-e lanes.

790FX

        * Codenamed RD790, final name revealed to be "AMD 790FX chipset"
        * Dual-socket (Quad FX, Dual Socket Direct Connect Architecture) or single AMD processor configuration
        * Maximum four physical PCI-E x16 slots and discrete PCI-E x4 slot , the chipset provides a total of 52 PCI-E lanes, with 41 lanes in Northbridge
        * HyperTransport 3.0 with support for HTX slots and PCI Express 2.0
        * ATI CrossFire X, see below
        * AutoXpress, see below
        * Extreme overclocking, reported to have achieved about 420 MHz bus for overclocking an Athlon 64 FX FX-62 processor, from originally 200 MHz.
        * Discrete chipset cache memory of at least 16 KB to reduce the latencies and increase the bandwidth
        * Supports Dual Gigabit Ethernet, and teaming option
        * Reference board codenamed "Wahoo" for dual-processor system reference design board with three physical PCI-E x16 slots, and "HammerHead" for single-socket system reference design board with four physical PCI-E x16 slots, also notable was the reference boards includes two ATA ports and only four SATA 3.0 Gbit/s ports (as being paired with SB600 southbridge), but the final product with SB700 or SB750 southbridge (see below) should support up to six SATA ports
        * Northbridge made on 65 nm process, manufactured by TSMC, and runs at 3 W when idle, and maximum 10 W under load, nominal 8 W power consumption, the northbridge was seen on reference design with single passive cooling heatsink only instead of connecting to heat pipes which are frequently used on current mainstream motherboard offers, the combination of 790FX northbridge with SB600 southbridge consumes normally less than 15 W
        * Enthusiast discrete multi-graphics segment

Even if the Intel system is faster the amd system with less costly MB and much cheaper ram will likely be a better buy.

Re:MHz wars are over

[Firethorn]

This sounds a lot like a '640k' quote to me.

A properly functioning word processor can already do pretty much everything 99.99% of what a user asks of it as fast as the user can tell it to do something, even on the bottom line processor.

Today's video games, sure, aren't going to benefit much from multicore. But I disagree that the benefits for future games will top out at 2. I mean - you could have 1 core handling user input and processing, 1 core handling the physics enviroment, 1 core for unit AI, 1 core for graphics information. There's a quad core right there.

Business and scientific apps will see some beyond that, but memory tends to be the bottleneck there- we'd be better off increasing memory bandwidth and latency than clock speed

Then they can start worrying about beefing up memory bandwidth - I've read about some technologies in the pipe that will help with this. And the scientific community can always use more bandwidth - they are one of the larger users of supercomputers, and this might take a project from 'Need to rent 24hrs on the supercomputer for $$$' to 'I can run this on my work computer for a month/week to get the same results for $'.