AMD snags Guinness World Record for clockiest chip The overclocking scamps of "Team AMD FX" are celebrating the Intel Developer Forum's opening day in their own special way: by accepting a fastest-chip-ever award at an event a mere block away from Intel's geekfest. "We applaud AMD for their entry into Guinness World Records for achieving the Highest Frequency of a Computer …
Time and hardware? are you serious? A liquid cooled cpu block and chiller would do this off the shelf from newegg for $100 or so. Considering they were looking to break a worldwide record (read faster that Intel or anyone else) they probably used Liquid nitrogen. Which again would cost only a few $100 at most. and Pointless? Well the point is they are showing their 8 (yes eight!) core cpu is likely to sell Retail clocked somewhere in the 4-5Ghz range. Any college student would do this work for the demo for free, or just have a few employees spend the afternoon slapping it together. This also means with aftermarket coolers, and normal liquid chillers, home users could snap up an 8 core CPU running in the 5.5 to 7Ghz range for only $40-100 extra in cooling and still be on par in cost with an Intel retail chip.
Can retail AMD chips do 4Ghz? They already do! the 2-3 year old 945 Phenom/Athlon II could reliably clock between 3.8 and 4.2GHz sometimes even higher with just a $30-$40 aftermarket cooler (and just one with nice heatpipes and good fan mind you, no water cooling or liquid nitrogen needed.)
Don't forget that AMD chips did and still outperforms similarly priced Intel Chips..."bedraggled" indeed. AMD has kept up their market share gaining 1.5% while intel lost 1.5%, and actually made good improvements into the server market as well, taking more share from Intel in recent times:
From the Reg just a month ago:
I"n terms of unit shipments, AMD gained a little market share thanks to the Fusion APUs, with 20.4 per cent of total PC microprocessor shipments in the second quarter, a gain of 1.5 points of market share compared to the first quarter. Intel had a commanding 79.3 per cent share of the PC microprocessor pie, and lost that 1.5 points of share."
http://www.theregister.co.uk/2011/08/02/pc_microprocessor_q2_2011/
Not to mention AMD did this all despite Intel's very illegal monopoly tactics and dirty business ethics, and a constant overlooking by the most of the press like Intel fav's Toms Hardware (though to be fair Tom's has been choosing AMD chips for their budget gaming systems quite a lot).
And you can buy a 945/955 CPU now for about $100 or less, and I bet if you have an AM3 motherboard you will be able to drop one of these nice new 5+Ghz processors without having to buy a new motherboard, let alone a whole new system. USB 3.0, oh yeah AMD Motherboards have been supporting that on most models for over a year now, so if you have a recent PC, no need to upgrade MB at all.
But go ahead and snap up that new Core i7 for $700 if you care too, you will loose respect of economical folks like myself, but hey you can have FPS bragging rights..at least till AMD's new chips come out :-) The AMD Fusion chips hold an awful lot of promise too especially for gaming laptops and budget gaming PC's Or maybe you like the integrated video Sandy bridge offers...didn't know that it had video, oh that's right its so bad even Intel isn't promoting that feature.
(Oh and yes this cheap 945 CPU with a simple BIOS change to overclock it will run Crysis with all the bells and whistles... at least if you pair it with a nice new AMD video card anyways)
The reason is even though the manufacturers have got better at it, they still bin their chips based on simple tests using cooling equal to that of their stock coolers. even a $20-30 aftermarket cooler can let you get a 3.4 or 3.5 Ghz CPU up to 4.0 Ghz without anything more than a simple BIOS change. And you get a better chip and get to pay less money for it than buying the highest clock version available.
Intel doesn't let you do this because they lock the CPU multiplier, so you have to crank up the frequency of the Bus (and memory, PCI etc) which doesn't work out so well.
Intel does this to maximize their profit potential and is part of why AMD is great! not only for us lucky AMD users, but for the CPU market as a whole, because it keeps Intel from raping their customers (well at least not quite so bad) on price.
And yes games now come standard with multiple graphics options, so its not a question of a game working or not, but with the CPU cranked up to its full potential you can afford to turn on nice features like 16x AA and very high texture details that really make games look more realistic, or just run the same maxed out details with smoother FPS. Also with 8 cores, clocking up from a theoretical retail 4.5Ghz to 5.5Ghz will help with other CPU intensive tasks like video rendering, or multitasking (Anti-virus and the like in the background). Sadly it takes me an 3-6 hours on my old media PC to turn a few Video .avi files into a DVD movie (DVDflick and ManDVD are pretty cool programs). I can't imagine how snappy a 5+GHz 8 core CPU will be in comparison to my antique single core 3.2Ghz CPU (2.4 CPU not overclocked btw)
for my T5750 @2.0 Ghz, dvd encoding is i/o bound.
upgrading to a faster processor (and required mobo + ram) will do bugger all
(except for making faster busses available ;) )
For me, a faster HD will have a much bigger impact, at least until HD speed is fast enough for it to become CPU bound again.
Am currently leaning toward an APU with a nippy SSD...
"dvd encoding is i/o bound"
As is virus scanning, in all likelihood - another task mentioned by the OP.
I know that with the workloads I have - mostly software development, including some largish builds and test runs - I/O performance is vastly more important than CPU performance, even with software bloat and eye candy.
Obviously there are many CPU-bound workloads, but I suspect for most PC users overclocking is irrelevant.
Certainly Disk IO is a factor, but even though its Old I built the system smart.
Media is read from one pair of stripped IDE drives, and written to a second pair of stripped SATA drives. OS and swap are on a RAID SCSI 3 disk. So disk I/O with my sad single core CPU is not the bottleneck. This work is especially more CPU bound as my old CPU has very little on die cache compared to more modern chips, and it really shows when the DVD encoding is changing the file format (having to re-render every frame of the video file).
I have a dual core laptop (with single 5400rpm drive that will encode a DVD in LESS than half the time the media PC takes. No granted I haven't this not even comparison as the laptop encoding was done using ManDVD (on Linux) and the Media PC is using DVDFlick (on XP). So if was so I/O bound the laptop would be coming in WAY behind the PC with its multiple striped disk arrays (they do about 66/MB a second data transfer on a basic disk speed test)
Just check out a site like Tom's to charts and see how much faster more modern CPU's are capable of rendering this work.
Granted a lot of this speed increase may come from faster buses, but that is part of what new CPU's and MB bring to the table
I have a dual CPU quad core Xeon server that can slice through these files in minutes instead of hours, and that is using a single SATA drive for the OS and filesystem (a test setup, I know it sounds dumb) Only problem is it sounds like 10 hairdryers going on full all at once.
So from my experience CPU seems to be a much bigger factor that disk I/O.
Your media PC is better specced i/o wise than my main system, so in your specific case, with all those disks, an older CPU, you are correct. However, I doubt that many people have that sort of setup, most NAS devices tend to be set up as either mirrors, JBOD or RAID5.
If people design the system to fit a specific task (as you have done), they can target the bottlenecks, and not waste money on new/overspecced equipment that wont benefit the job in hand. However, in general, when buying a ready made PC it is always the I/O that shows its age first (namely hard drive thrashing with too little ram). This is arguably down to software bloat (I'm looking at you Firefox :-|). Outside of games, there simply aren't that many things that are compute bound. At least not for the home user, and everyone else knows enough to tailor their machine to its task.
Id be interested to see the CPU load on the Xeon test you ran...
Atom processors and arm devices have taken off because these chips do almost everything the home user wants to do, despite being *much* slower than what is mentioned in this story. The rise of virtualisation came about partially due to how underutilised the contemporary server hardware was. BOINC and it's ilk are unobtrusive because they use very little I/O (as far as Seti/folding go). I could go on...
IIRC from the Toms Hardware project, the motherboard power components have to be upgraded, and other components also need to be refrigerated. Plus there's the condensation to deal with, etc. The old Crays were bathed in nonconductive coolant. It would be easier to have a sealed system with the motherboard in a Freon bath.
The article says: "will offer completely unlocked processor clock multipliers for easier PC enthusiast overclocking ... giving overclockers and PC enthusiasts complete customization and flexibility.". What this does is allow normal "clocktards" the ability to see just how far they can overclock a processor chip. When they go to far and burn up a chip (they do generate heat!) AMD will gladly SELL them another chip.
Wow, another sale! What great margins. AMD stock soars.
As another said, this will probably be the minimum necessary for Windows 8!
Still a damned good thing that AMD is around, and if you want to read about the Intel/AMD saga borrow "Inside Intel".
The sad thing about Intel and AMD is that their crippled design managed to outdo RISC, a much better design.
It would not surprise me, nor make me happy perhaps, if the Chinese will manage to change that as they apparently have adapted RISC as their design.
It's not clear that a RISC processor is better at all things than a CISC, even after having been around for 30 years or so.
That's why modern RISC designs like POWER, SPARC, and even the good old ARM processor are having complex instructions added to their ISA (such as Thumb 2, VFD and NEON) as time goes by.
Increasingly, the difference between an augmented RISC processor and a CISC processor with some of their frequently used instructions being engineered to run in small numbers of clock cycles is becoming more and more different to see. It now appears to revolve around electrical power rather than computing power.
But it's all irrelevant, really. On a personal computer, unless you do hard-core gaming or real-time media transcoding, you just don't need anything much faster than around a 2GHz processor with some graphic assist. We've just got so used to bloated OS and application code that we accept that ever-faster processors are required without questioning why we need them.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
