Core i7 launch
Posted: Mon Nov 03, 2008 10:48 am
Discussions about Silent Computing
https://silentpcreview.com/forums/
https://silentpcreview.com/forums/viewtopic.php?f=13&t=50831
Wow, this was completely out of place... I don't want to start explaining how different things are nowadays, let's just say that simply because a CPU doesn't fit your needs isn't a reason to call it a failure. I wonder what you mean by "something for desktops"?..QuietOC wrote:I think this may be as big a failure as the Pentium 4
No, it does not. It has very high idle power, which is what most desktop CPUs do most of the time.npp wrote:And what is wrong with the Core i7, anyway? It has an even better performance/watt ratio than current Penryns, and is architecturaly superior to them in every aspect.
Not that THG is a great source to quote, but they measured idle power to be some 1-2W in idle. It's load power consumtion that has increased, but given the higher performance you get, it actually transfers to lower total power consumption after all, as seen @techreport. I'm curious what anandtech is going to some up with in the next days, Anand seems to be working on a power consumption analysis right now.QuietOC wrote:No, it does not. It has very high idle power, which is what most desktop CPUs do most of the time.
I don't see any reason to want one of these at all, especially a 965- not for HPC, not for home computing, and definitely not for business use. With the G7's economies going the direction they are as of late, I don't even see where this will sell well to the narrow niche of spoiled children.QuietOC wrote:I think this may be as big a failure as the Pentium 4.
Some existing Core2 products will be discontinued in 2009, but other new Core2 products will be launched. Do you think this contradicts or supports my statement about Core2 products being widely available for at least 2 more years?npp wrote: @jessekopelman: check out the article here: http://www.techpowerup.com/71659/Intel_ ... ineup.html
I hadn't read Tom's. Well, there might be some hope for Intel desktop chips. AMD's Phenoms also have horrible idle power consumption, and I wonder if high speed QPI/HT links are to blame. Power is proportional to frequency ^ 3, and both AMD and Intel are now both high clockspeed buses compared to the original 800 MHz HT speed.npp wrote:Not that THG is a great source to quote, but they measured idle power to be some 1-2W in idle.
The E8xxx series was released less than year ago (in January), and it's about to be phased out now. You should be able to do the math for yourself. I just wanted to point out that another 2 years are an unrealistic lifespan for current Core 2 CPUs (just as it is nearly impossible to get a Pentium D nowadays - not that anyone would want to do that).jessekopelman wrote:Some existing Core2 products will be discontinued in 2009, but other new Core2 products will be launched. Do you think this contradicts or supports my statement about Core2 products being widely available for at least 2 more years?
Won't it just be renamed to Celeron something, and live like that for a while?npp wrote: The E8xxx series was released less than year ago (in January), and it's about to be phased out now.
I think you are misreading the article you linked. Specifically the part where it says, "all products will phase out between the first and second quarters next year." All products is all products listed in the PDNs, not all existing Core2 products. Only a single E8XXX is specifically listed as being discontinued, the E8300, not the entire line. Also, the article lists a gaggle of chips launching on 11/30 and a E7500 that will be launched in Jan 09. You think these products will all be discontinued < 6 months after launch? Even if so, end of production does not mean end of availability. You can still get a brand new Prescott P4 from Newegg! Anyway, as others have stated Core2 technology is bound to live on as Celerons or the like for quite a while. They were still making Celerons based on Core up until a few months ago (maybe still are for the OEM market). This issue would be getting a lot more press if your take on it were the norm.npp wrote: The E8xxx series was released less than year ago (in January), and it's about to be phased out now. You should be able to do the math for yourself. I just wanted to point out that another 2 years are an unrealistic lifespan for current Core 2 CPUs (just as it is nearly impossible to get a Pentium D nowadays - not that anyone would want to do that).
You're not the target market for these chips. Intel doesn't care that you're nonplused and unwilling to spend ~$,1000 on one of these new CPUs. Intel is also well-aware that plenty of people can do everything they need with a Pentium M. For those of us running applications that thread well, though, and need lots of bandwidth, it's going to be a beast. There are things at work that I wouldn't even bother trying to run on my current workstation that Nehalem will eat up. How's that for a reason to want in on one of these? Going from 'not able to do it' to 'able to' is as big a performance boost as you can get.fri2219 wrote:I don't see any reason to want one of these at all, especially a 965- not for HPC, not for home computing, and definitely not for business use. With the G7's economies going the direction they are as of late, I don't even see where this will sell well to the narrow niche of spoiled children.
Some of the market is well-served by Atom-powered boxes. And what novel CPU tech has provided a ten-fold performance leap over the last generation?When the chip market is moving to Atom-sized margins and good enough for web surfing performance, I can't see how a chip that isn't a ten-fold leap in dual-threaded performance is going to be a success. Anything more than two threads at a time seems to be beyond most application developers, even though CS programs been teaching non-blocking, multithreaded I/O, for two decades.
Everything about our lifestyles damages the environment. Your CRT isn't exactly innocent as the driven snow. It may be a sunk environmental cost now (not really, because some day it needs disposal), but it still has a footprint.npp wrote:In the grand scheme of things, 99% of the products launched around shouldn't concern me; I've got an old slow 2Ghz PC that will serve me well for the next time, and I really appreciate the damage everyone of us does to the environment when buying new products one actually don't need. I won't upgrade to Core i7, nor would I buy and LCD TV as long as my old CRT does the job well. For me, the Core i7 is interesting solely as a technological leap, not as something I would potentially buy.
You clearly know what I meant when I wrote this. I never said my CRT is "innocent as the driven snow", it's an example for a tech device that I have used for many years so its environmental cost gets amortized over time. If you want to get picky, you must kill yourself at the very moment, since almost everything you do damages the environment you live in. My point was that we should try to use the things we have as long as they would serve, and not just upgrade them because of "wow, my rig looks aged now, what do I do" or "god, those new 60" LCD screens look really nice in my living room".Woland wrote:Everything about our lifestyles damages the environment. Your CRT isn't exactly innocent as the driven snow. It may be a sunk environmental cost now (not really, because some day it needs disposal), but it still has a footprint.
I've only skimmed a couple of articles so far so maybe this has been covered already, but hopefully the people doing the power testing recognize that the memory controller is now on chip, so you really need to take the whole CPU/northbridge/memory subsystem into account when comparing power consumption to the Core 2 (could be problematic given the large variety of Core 2 chipsets with widely varying efficiencies). So it could be that the power savings are even larger when you consider how much power some of the Core 2 northbridges draw, and also that you now have one fewer power-consuming I/O bus involved in a memory transaction (specifically, the CPU-to-NB interconnect).npp wrote:Not that THG is a great source to quote, but they measured idle power to be some 1-2W in idle. It's load power consumtion that has increased, but given the higher performance you get, it actually transfers to lower total power consumption after all, as seen @techreport. I'm curious what anandtech is going to some up with in the next days, Anand seems to be working on a power consumption analysis right now.QuietOC wrote:No, it does not. It has very high idle power, which is what most desktop CPUs do most of the time.
@jessekopelman: check out the article here: http://www.techpowerup.com/71659/Intel_ ... ineup.html
It seems the new chipset is a huge power consumer. There are no fewer total busses with just moving the memory controller from one chip to another.Metaluna wrote:So it could be that the power savings are even larger when you consider how much power some of the Core 2 northbridges draw, and also that you now have one fewer power-consuming I/O bus involved in a memory transaction (specifically, the CPU-to-NB interconnect).
Yes the bus is still there, but I was thinking more about bus utilization. You would still use that bus for PCIe traffic, for example, but for memory intensive operations it no longer needs to be involved which should in theory save some power since off-chip buses tend to use much more power than an on-chip connection between the CPU and memory controller. Of course, if the NB turns out to be a power hog anyway then the point is moot. It sounds like the X58 is targeted towards people running systems with SLI setups that idle at 100+ watts anyway, so they probably won't care. Hopefully we'll see some better options eventually.QuietOC wrote:It seems the new chipset is a huge power consumer. There are no fewer total busses with just moving the memory controller from one chip to another.
This is not a DataCenter chip- i7 doesn't compare favorably to the T2 or Power6 for data center applications. Several Intel speakers said as much at the last IDF sessions aimed at scientific computing, and the engineers I know at the Ronler Acres fab have said the same as well (off the record, of course).AZBrandon wrote:This continues to make the case that they designed the chip with datacenters and blade computing in mind
I think that might depend on whos data centre you're talking about and what it's doing. The term seems to be used very loosely to apply to any room with more than a dozen servers in it!fri2219 wrote:This is not a DataCenter chip- i7 doesn't compare favorably to the T2 or Power6 for data center applications.
Well if you're in the market for a T2 or POWER6 I'd bet that Intel would rather sell you an nice expensive Itanium rather than a cheapo Nehalem. It's not surprising that Intel would downplay the relevant of Nehalem in such a situation.Several Intel speakers said as much at the last IDF sessions aimed at scientific computing, and the engineers I know at the Ronler Acres fab have said the same as well (off the record, of course).