What is Digg?72 Comments
- Muncher, on 10/11/2007, -6/+27"'Our true hope is to reclaim digg user Muncher's avatar,' Advanced Micro Devices CEO Héctor Ruiz said in a press conference on Wednesday. 'He has often expressed regrets about betraying AMD, and we hope that Barcelona will cause him to reverse that decision. The truth is, he has singlehandedly caused demand for our Athlon product line to dramatically decrease. We believe that swaying his allegiances, by releasing a truly superior product, will the the key to a strong third quarter.' Muncher was unavailable for comment."
- Nobiting, on 10/11/2007, -3/+18That's a pretty ironic avatar you've got there ;)
- kingkilr, on 10/11/2007, -1/+12On July 22nd the Q6600(2.4GHz quad core CPU) goes down to $266.
- Reziarfg, on 10/11/2007, -6/+16Finally! I've been an AMD person for quite a while, but my last two machines have been Intel because of their lead on AMD lately. It's good to hear AMD is back in the game. Anyone know whatever happened to the 4x4? I remember reading something about it, but it never really hit it big.
- Reziarfg, on 10/11/2007, -2/+11Look closer ;)
It's a bit outdated though - inactive, on 10/11/2007, -4/+11Once I find a use for dual-core, I might be tempted by quad-core.
- EntangledPhysx, on 10/11/2007, -0/+6Well, are you going to let us in on whats happened on the 22nd of July?
- jonnyboy88, on 10/11/2007, -0/+6how about waiting till it's actually out and usable before judging it?
- geminitojanus, on 10/11/2007, -0/+6Fact of the matter is "True Quad Core" as a marketing term has completely failed AMD. Intel has maximized the quad core market, and they did it way more efficiently (by instead of engineering the more complex solution, grabbing two of their already fast dual core chips and slapping them into the same package).
Even AMD admits that if they could go back in time they'd go with a MCM solution; they're easier to engineer, faster to market, and AMD would still have the Coherent HTT bus on their side, even if it was more latent from core to core than being on one piece of silicon. - neocr0n, on 10/11/2007, -0/+5Price drop on Intel as far as I know.
- foxhoundadmin, on 10/11/2007, -4/+9you know what's sad? you got dugg down, because more than half the people on digg don't know why july 22nd is important.
take THIS comment, for example:
"What happens on the 22nd of July? Is it all your birthdays? LOL"
sarcasm? probably not. just your typical digg user who doesn't know jack about technology and/or tech news. - ScrabbyDoo, on 10/11/2007, -0/+4Needless to say, what you just said was needless to be said.
- imwithjesus, on 10/11/2007, -19/+23tempting, but i'm still building my pc around july 22
- MrObjectional, on 10/11/2007, -3/+7The entire Core 2 line drops in price on July 22.
- Chandon, on 10/11/2007, -0/+34x4 is actually really nice - it's noticably faster than Intel stuff if you're doing anything memory-limited. The big problem is that most of the mainstream applications go through contortions to not be memory limited (and not to take advantage of NUMA), and therefore 4x4 doesn't help you on most commercial desktop software workloads.
- MoFoKeR, on 10/11/2007, -1/+4So on july 22 when the core duo drops price wont amd have another price drop to? Just to counter intel..If thats the case ill go for the top of the line amd just becouse I can still get the extra processor which no program or better yet any game has put to good use to date...Same will go when those 4 processor cpu's come out..Why waste the extra money thats ..BEER MONEY !
- EntangledPhysx, on 10/11/2007, -2/+5it's your momma's birthday
- Scheitbag, on 10/11/2007, -2/+5How about some software that truly pushes these multi-core processors....
- AlexFerny, on 10/11/2007, -1/+4not thier fault software developers are to crap to write multi threaded things
- Chandon, on 10/11/2007, -0/+2The difference between DDR2 and DDR3 is pretty small compared to the bandwidth gain from 4-way NUMA with on-chip memory controllers. The Xeons are neat, but for a 4 socket server running some memory-limited task where you can use NUMA, they really can't even touch Opterons.
- extrasalty, on 10/11/2007, -0/+2I'm tired of stupid remarks that Intel will "kill" AMD, that the ATI merge will kill AMD, that Penryn will kill everything. I've had so much I had to register and give my honest opinion.
AMD realized a while back that you can't get rich on making CPUs alone- example Intel. As the famous quote goes, nowadays CPU fans cost more than the cpu itself. The move from $60 part (Sempron) to $400 (2900XT) is what will make Intel fanboys, like the ones here, spent more on AMD parts than on Intel- don't say you wont- just wait for 65nm 2900 w/ 1GB. The only thing more insulting to Intel is for AMD to make a mainstream chipset for C2D. The ATI/AMD merge is so important for the industry the Intel started trumpeting Larabee.
I used to overclock a lot- my mobile athlon that I got for $85 OCed from 1.8GHz to 2.35GHz- admit it- it is exciting to get 30% OC, but really if you play games, a video card does a lot more for you- hence why I chose the cheapest CPU. With the recent CPU price cuts OC is a thing of the past for me- my 1950Pro (AGP never-the-less) and x2 5600 do 5000 3dmark 2006 and 10000 on 2005- that's perfectly fine for a $45 MSI (enough said) motherboard and $90 2Gb.
Who cares how fast Penryn or Neh-who will be- FACT: most applications are not designed for multicore processors, and the ones that are, cost a hefty buck because they are professional applications and time is money. It also happens to be that new technology CPUs are close to $1000. So top performance is for top players. Who cares how fast of a processor you have when XviD does 75% utilization- it's still inefficient. Intel will still spend a lot of money convincing you that you need that power. Now just imagine what will Intel have to do to counter $300 native(cheaper to make) quad core and how much will that eat into their margins both in home and server market.FACT: most home computers idle majority of the time - it sure helps to spend heaps of money for that stellar performance. I think just recently after seeing HP's success, Dell realized the "Wal-mart" potential- customers going for the lowest price regardless of brand or performance- reason- consumer appliances have been around for a while and people are getting used to having a wide choice of prices with the lowest being most popular. AMD has been the king of "cheaper than Intel" and is the only logical choice for rock bottom computer price...it happens that AMD makes video cards too...hmmm let's throw 1, 2 or 3 of them in this brand spanking Dell and sell it in Wal-mart (in front the heart of Wal-mart, right next to electronics department) and see what happens.
Time will tell the reasons for the delays with Barcelona but I wouldn't be surprised when Dell comes out in August (back to school) loaded with them.
Have a great life! - Chandon, on 10/11/2007, -0/+2Programmers have gotten a free ride for a long time with continually increasing clock speeds on fully-backwards-compatible chips. Those free performance gains have stopped, unless people want to write innately parallel code. If they do, they'll get the performance gains again for a while. End of story.
- Chandon, on 10/11/2007, -0/+1Sure, sequential programming has worked for a long time, has been the most efficient solution, and has gotten free performance improvements as clock speeds increased. Programming in an innately concurrent style has worked for a long time too, and has been slightly less efficient than sequential code. Now innately concurrent code gets the free performance boost and sequential programming doesn't... it's time for people writing performance-critical code to learn a new style.
- Chandon, on 10/11/2007, -0/+1Nope. The reality is closer to this: The world is naturally parallel, and so software that interacts with or simulates the world is also naturally parallel. The problem is that the development tools and abstractions that are widely used today favor sequential code, and most programmers are used to writing sequential code. As the technical requirement to write parallel code becomes more obvious, people will get their act together and learn to do it.
- Xentrion, on 10/11/2007, -0/+1I hope this processor is compatible with Socket AM2, changing sockets would be a huge pain.
- Arramol, on 10/11/2007, -1/+2"Anyone know whatever happened to the 4x4?"
It lost a lot of its appeal after the QX6700 and 6800 came out. - LordofShadows, on 10/11/2007, -0/+1Even if you had a new programming language that allowed you to achieve perfect parallelism within a single program you would still not be able to fully use the amount of cpu cores the hardware guys would like to throw at us. -- What do you mean the software isnt faster? We gave you 10x the cores, should be 10x as fast right?
After they give us too many cpu cores mfgs will look at add other whizbang features: gpu's, vpu's, any other feature not related to raw single core speed bumps. -- All the while blaming software, cause software fixes everything... - geminitojanus, on 10/11/2007, -0/+1Once again, it's hilarious to see how many people don't realize that every major chipmaker on the planet has released errata for their CPUs. Nobody gets it perfect the first time around. In fact, AMD's errata sheets seem to be longer than Intel's, and Intel's errata for the Core 2 is full of things like "This is because we changed something between Core and Core 2, if you understand how it's changed it's no problem for you".
These recent TLB changes are the biggest thing that showcase this, and are a complete and total non-issue (something a Microsoft OS engineer would spend a few hours and a cup of coffee working on). If you can honestly put the processor into that failure state without an obtuse test case, you're pretty damn skilled. - HsoKinees, on 10/11/2007, -1/+2interesting.. i've never liked or trusted CNET, and this just adds to it.. immediately after going to the article, i won a virus prompt! must be from one of their advertisers..
http://image.bayimg.com/na/co/ia/ab/o.jpg - darkNiGHTS, on 10/11/2007, -0/+1Funny, but why are people like this? Isn't it natural that when AMD gets ahead for a while, Intel releases their new technology and gets ahead? AMD then has to develop new technology to beat Intel's technology, which takes a while.
- darkNiGHTS, on 10/11/2007, -0/+1wrong reply
- geminitojanus, on 10/11/2007, -0/+1They've been possible for at least a decade, the term "Very Long Instruction Word" (some graphics cards) and "Explicitly Parallel Instruction Computer" (Itanium) bounce to mind immediately. These processors are absolute monsters of math, but they're much harder to program due to the need to use instruction packing, which is a concept most compiler designers aren't used to (as pretty much every other computer they'll ever use in their entire life is SISD with a little SIMD and no MIMD).
In the meantime, we can just go on programming like we have been; 98% of applications _DO NOT NEED_ to be multi-threaded, and the ones that do need to be threaded already are, or have people clamoring over the right now to make them so. Remember, we've lived for ages with the fact that we only have one processor and we need to split everything out into individual timeslots, now we have several processors and the granularity of those time slots can be reduced or at least reorganized to better fit the whole system. Odds are you won't need to do anything special years in the future to run your app on a multi-core machine, just keep coding like you're used to. - inactive, on 10/11/2007, -1/+2*waits on july 22* newegg is just waiting on me!
- Chandon, on 10/11/2007, -0/+1Why? AMD's socket changes are the single most overhyped thing in internet discussions - 90% of the time when you buy a new processor you buy a new motherboard to go with it, even when you are building yourself.
- inactive, on 10/11/2007, -0/+1Yes, but in the larger scheme of things, your going to find a lot more Xeons being used out there as opposed to the Opterons. The memory speed was the issue they didn't work on early enough. Only now do they support DDR2 and myself personally get a bottleneck at around 2gb. My DDR1 could not feed the processors fast enough. It is probably still the case with DDR2 and the new sockets. Hopefully DDR3 will begin to take advantage more.
- harshbarj, on 10/11/2007, -0/+1Amd is also having a price drop soon. So we will have to wait and see if the q6600 is really the best chip for the money. As we have already seen the latest Amd price cuts made Amd the chip to have for budget systems (the x2 3600+ is hard to beat at ~$60 as Intel only has the Pentium d in that price range).
- Hangender, on 10/11/2007, -1/+2Who cares. The only good will be that Intel will also drop its price on quad core CPUs to compete with AMD. The customer wins
- caleb4mj, on 10/11/2007, -0/+0You would pay $400 for an ATI 2900XT? I would buy an nVidia today, and hopefully tomorrow I will have a choice of open source graphics from Intel. If that's AMD's plan I feel sorry for them.
ATI was a bad decision, I could have told them that before the purchase. The only way they can fix it is to open source the products and ask the community to help spread the word. As of now I'm spreading the word that Intel has open source graphics drivers, that is you can get the source code to their unified shader chip so you can use the hardware for whatever you can think up, not just what ATI provides for you in Vista.
And I will continue to repeat this until AMD "gets it". FACT: I'm not buying a multicore processor for "most applications" that are not designed for it. I'm buying it for price/performance and multicore applications, which I'm perfectly capable of writing myself. I know my hardware intimately. If AMD wants me to recommend their hardware isn't really simple, just support Linux as well or better than Intel. I'll find a reason to recommend AMD, if they give me a reason to look again. - thegreatanti, on 10/11/2007, -0/+0You meant: Sell your AMD stock. Now it has some value, next year it won't.
- thegreatanti, on 10/11/2007, -0/+0AMD will be crushed by Intel's Penryn. Period.
- Chandon, on 10/11/2007, -1/+1If making a "Massive single core CPU" was feasible, both manufacturers would be all over it. App developers will just have to get their acts together and write for SMP.
- DestroyFascism, on 10/11/2007, -4/+4I take it this is a true quad core?
- Chandon, on 10/11/2007, -1/+1Lots of people are still buying Opterons, and for far more money than the Xeons, because servers - and expecially compute clusters - care about memory bandwidth a lot.
- mickrussom, on 10/11/2007, -0/+0http://www.reghardware.co.uk/2007/06/29/intel_2ghz_clovertownlv/
- Chandon, on 10/11/2007, -1/+1The problem isn't that AMD is wrong about "Native Quad Core", it's that the benchmarks are largely designed for multi-socket servers (which are more like Intel's MCM design than AMD's design). Code that actually communicates between threads will go much faster on a "Native Quad Core" processor - communication is necessary in a number of efficient parallel algorithms, so we should see benchmarks that take advantage of it as soon as there is a processor to run them on.
- fLUx1337, on 10/11/2007, -8/+8"tempting, but i'm still building my pc around july 22"
Same here, July 22th is when the Q6600 Core 2 Quad is roumored to be lowered to $233, which will be around £150 for me....
So really, stuff AMD, I'm no intel fanboy, but I'm not waiting months just to see what they have to offer.... - markey, on 10/11/2007, -2/+2Haha :)
- mhmdkhamis, on 12/14/2007, -0/+0http://game.paramegsoft.com/
Who cares how fast Penryn or Neh-who will be- FACT: most applications are not designed for multicore processors, and the ones that are, cost a hefty buck because they are professional applications and time is money. It also happens to be that new technology CPUs are close to $1000. So top performance is for top players. Who cares how fast of a processor you have when XviD does 75% utilization- it's still inefficient. Intel will still spend a lot of money convincing you that you need that power. Now just imagine what will Intel have to do to counter $300 native(cheaper to make) quad core and how much will that eat into their margins both in home and server market.FACT: most home computers idle majority of the time - it sure helps to spend heaps of money for that stellar performance. I think just recently after seeing HP's success, Dell realized the "Wal-mart" potential- customers going for the lowest price regardless of brand or performance- reason- consumer appliances have been around for a while and people are getting used to having a wide choice of prices with the lowest being most popular. AMD has been the king of "cheaper than Intel" and is the only logical choice for rock bottom computer price...it happens that AMD makes video cards too...hmmm let's throw 1, 2 or 3 of them in this brand spanking Dell and sell it in Wal-mart (in front the heart of Wal-mart, right next to electronics department) and see what happens.
Time will tell the reasons for the delays with Barcelona but I wouldn't be surprised when Dell comes out in August (back to school) loaded with them.
http://www.paramegsoft.com/forum/ - geminitojanus, on 10/11/2007, -2/+2"Of course power consumption is another thing entirely and should factor in accordingly."
Judging on all pre-release benchmarks currently available, the Barcelona is already in major trouble. Not only does it not perform clock-to-clock as well as current Core 2 processors, it's IPC is lower (in case you don't know what this means, to put it in simple terms, the Core 2 can issue and retire more instructions per clock than the Barcelona can). Couple this with the low clockspeeds at launch, Intel's upcoming anti-Barcelona price cuts, and Penryn's November release, AMD is in trouble.
They had better hope they've fixed all of the bugs in the B0 stepping, else it's going to be a very long year-out for AMD. - abaker82, on 06/15/2009, -1/+1Lern sum PUNK_TU-A_shon
-
Show 51 - 71 of 71 discussions
Browsing Digg on your phone just got easier with our enhancements to the 
© Digg Inc. 2009
— Content created and posted by Digg users is 