Any word on that 8 core phenom i heard of a while back ?

Cause that would be sick for supreme commander

Cause that would be sick for supreme commander

brayant321

"With AMD pushing for 45nm later this year, I would expect late Q4 or early Q1." - LordEC911

Even at 45nm the die's would be a bit to large (mainly for heat problems) for an 8 core CPU. Also, AMD (nor anybody else for that matter) has not announced that they are working on a new chipset that can support a single 8 core CPU. Yes, you can currently run two quad's on some chipsets, but 8 cores on a single die is a bit different. Not to mention the fact that 8 cores on a single die would require quite a bit of cache sharing. Enough so that putting 8 cores on a single die would not be as efficient as two quad core CPU's.

I really wouldnt expect 8 core consumer CPU's for at least 3-4 years from now. Thats also assuming that we can get the die sze down to at least 30nm or lower. Thats a pretty tall order within that time frame. On top of all of that, there is very little reason for the average consumer to need 8 cores at the momment. And for those that do, well get yourself a dual, quad core opteron baord and go to town. I think you'll find that games like sup-com simply dont fair any better with 8 cores than with 4. * cores or more are really for very special tasks at the momment. Not games.

Even at 45nm the die's would be a bit to large (mainly for heat problems) for an 8 core CPU. Also, AMD (nor anybody else for that matter) has not announced that they are working on a new chipset that can support a single 8 core CPU. Yes, you can currently run two quad's on some chipsets, but 8 cores on a single die is a bit different. Not to mention the fact that 8 cores on a single die would require quite a bit of cache sharing. Enough so that putting 8 cores on a single die would not be as efficient as two quad core CPU's.

I really wouldnt expect 8 core consumer CPU's for at least 3-4 years from now. Thats also assuming that we can get the die sze down to at least 30nm or lower. Thats a pretty tall order within that time frame. On top of all of that, there is very little reason for the average consumer to need 8 cores at the momment. And for those that do, well get yourself a dual, quad core opteron baord and go to town. I think you'll find that games like sup-com simply dont fair any better with 8 cores than with 4. * cores or more are really for very special tasks at the momment. Not games.kodai

You might want to research before act like you know what you are talking about.

#1 The die would not be too large, nor too hot.
#2 Chipsets don't need to support more cores, in the way you might think they do. Plus both AMD and Intel have announced they are working on 8 core CPUs which, like I said, we will probably see in 2009.
#3 This is not 8 cores on a single die, AMD would be going the route of MCM, two dies connected thanks to HT3.0.
#4 What do you think L2 and L3 is for, cache sharing maybe?
#5 AMD can easily do octo core on 45nm, since they actually have less trannies in their quadcore vs Intel's. With the way Intel is talking about Nelahem, I would expect Intel to get to 32nm before they get an octo out.

I don't think average consumers are going to be spending the big bucks on enthusiast processors?
Do you?

Intel and AMD will rake in decent profits off of octocores, there is more markets out there then the personal and gaming markets... Enthusiast, server and workstation all come to mind and all three of those while not only raking in the majority of the profits will also make great use of octocore CPUs.

LordEC911, you know as well I anybody who bothered to ready my post that I was talking about conumer grade CPU's. I even said that in my post. Dont try to flame baiting me on this with stupid comments like "You might want to research before act like you know what you are talking about".

To produce 8 cores on the current 45nm die would require roughly 50% more space for cooling needs on the chip size iteslf. Given most PC's uses towers and most fansinks are starting at 1lb (some 3rd party fansinks are more that 2lbs), we are reaching a maximum load on weithc that boards can handle in the vertical position. Active air cooling is only going to go so far in cooling. The average consumer is not going to deal with liquid coolers, or heat exchangers. MB makers could start using reinfocements on the system boards, but given the cost and complexity that requires, I just dont see it happening. Instead, a major decrease in size of the CPU die will reduce the heat that a CPU gives off and therefore allow active air cooling to better seve the average consumer. That being the case, Intel has just finsihed upgrading to the 45nm process and AMD is just getting up to steam on this. The average lifespan on die sizes seems to be 3-5 years. If patters are anything to go by, it will be at least 3-4 more years before we see an 8 core consumer CPU. On top of all of that, there simply no demand for the consumer market to go for 8 cores. My comments about the cache were also about size relation. Hence I placed the comment in the post on dealing with size. It's a given that if you link multiple dies together you can have the same ammount of cache. Now lets take the q6600 as an example here. The ammount of room, power and heat related to say 16MB of L2 cache would make one wonder just how Intel would turn that into an 8 core setup. Simply put, they wouldnt. They would instead more than likley) make two quad core unites with each having between 4 and 6 MB of cache and then link those together to form an 8 core CPU. This means a design change, and that takes time as they have to make new templates and reqork the whole fab process for the new design. Neither AMD nor Intel has shown any sign that they are anywhere near being ready to do this on a large scale consumer process.

Yes, there will be server demand. Then again, that is not a typical consumer concern. I'm willing to bet that very few consumers have an OC24 line running to thier home for server needs to the general public. Or that many of them are working on the next "Deep Blue". Right now the real demand for quad core in the home is comming from those (like me) who work with vast ammounts of files in compression, and video editing. Outside of that, dual cores are currently filling the need that most consumers have. I'm willing to bet that Intel and (especially) AMD are not going to toss out hundreds of millions in design and fab plant upgrades to manufacture something on a mass scale (ie the consumer market) that wont sell because there is no real demand for it. Instead, they well continue there design work and start making them for industry work on a far more limited scale. Assuming there is no delay in thsi (yeah right), we may see them show up in the next year.

"I don't think average consumers are going to be spending the big bucks on enthusiast processors?
Do you?"

Well, no. Did you even bother to read my post. That was my entire point. I agree that AMD has a better chance of getting an 8 core CPU out before Intel does. This doesnt mean that it will be for the consumer market. It will be because AMD is trying to take another shot at the business market. Why do you think the bought ATI? They are gunning for the value an OEM has in the market place. ATI is the defacto standard intergrated graphics maker. AMD wants that to compliment there new push into lower cost workstations and high end servers. Meanwhile Intel is holding onto the lions share of mid range products and making headway into the intergrated graphics market. AMD's best chance for the high end server market is to produce a cost effective 6 or 8 core CPU that the average data cener can save money with. This does not translate into current consumer sales and therfore will not directly alter the consumer product line in the near future. Again, I still say it will be about two years before we see an 8 core for the business market and near four years before the consumer market gets one. And thats on the optomistic side of things.

If you want to continue berating other postesr on the forums, I would suggest you at least offer to speak via PM. To just blindly accuse people of not understanding what they are talking about and then telling somebody to do something like you have some authority over them is just plain rude and non conducive to the topic at hand. So if you want to discuss it further then PM me and we can talk. Otherwise I fear this thread will simply wander off topic and end up being locked while we both get moderated. I'd rather avoid that.

If you want to continue berating other postesr on the forums, I would suggest you at least offer to speak via PM. To just blindly accuse people of not understanding what they are talking about and then telling somebody to do something like you have some authority over them is just plain rude and non conducive to the topic at hand. So if you want to discuss it further then PM me and we can talk. Otherwise I fear this thread will simply wander off topic and end up being locked while we both get moderated. I'd rather avoid that.kodai

Well I call it like I see it...
Your rebuttal is no better than your first post.
Also trying to boss me around in your last paragraph is worse then me telling you to do some research.
Why are we not allowed to discuss things here? Don't whine and don't flame and we don't have a problem.

ATi is the "defacto standard of IGPs?" WRONG.
Intel is making "headway" in IGPs? WRONG.
Intel has dominated integrated graphics for quite a while. ATi has recently been slowly clawing back marketshare but Intel is still vastly dominating.

So your whole point for arguing with me that we won't see octocores is because motherboards can't support the extra weight of better coolers? Geez, grasping for straws are we?

Or your other point that Intel and AMD don't want to spend millions, which they have already spent, to get ready for 32/45nm respectively and eventually crank out some quadcores? (Yes quadcores, I will explain later)
Did you not read my post? Both of them are FAR into development of octocore. AMD has stated that their 45nm shrink will give them enough room to get it done, while Intel has always talked about both MCM and native octocore Nelahem.

Since the first octocore is going to be due to the MCM approach, this significantly DECREASES manufacturing cost per part while SLIGHTLY increasing cost due to packaging. Churning out a large number of quadcores will keep manufacturing cost down but the slightly more complicated packaging requirements shaves off some of that savings.

So I still stick to my first sentence in my first post. If you would take the time to read what they are working on, look at some roadmaps, read a few of their officer's presentations or even read some articles by analysts we could have avoided this whole mess. We have hit a wall with clockspeed and the current solution, until a whole new technology is found, is to throw more cores at the problem.

BTW- IMO Consumer market does mean the first desktop enthusiast processors, EX. QX9650 being released in '07.

Well, I see that you entire reason to post in this thead is to troll and get yourself and another modded. Maybe even clsoe it. This will be my final responce to you on this topic. First off i think the way you see it's is with blinders. Secondly I never made a "rebuttal" as I'm not arguing the your point. I'm trying to state that your point with my post in not realted to the topic at hand. The OP was clearly asking about consumer grade CPU's (as he brings up games in his orginal question), but you keep discussing non consumer based CPU's. Also, I nevertried to "boss" you arround. I politely suggested (hence used the word "suggest" in the line "I would suggest you at least offer to speak via PM."). I also NEVER said nor implied that we were not allowed to discuss "things here". Thats all you. I merly pointed out that the GS ToU requires us to stay on topic or risk the rath of the mods. I simply wanted to avoid this exact scenario as it sometimes causes threads to be locked and users to get a moderation. You know that is what I was refering too, and yet you continue to press the matter. Why? You are simply out to cause trouble as far as I can tell. Thats the last I'll dicuss on this matter. If you bring it up again, I'll just ask the mods to lock the thread and let the chips fall where the may.

Getting back to the topic of the thread I'll say this. Dell, being the largest workstation supplier in the world with it's optiplex line, has sold more optiplex's with ATI than Intel by a very large margin. The numbers are not that hard to find. HP/Compaq are close behind. Simply put there are more ATI based workstations in the business world (and that makes up the majority of active computer in the world) than Intel. Hence ATI is the leader in this market. As I said though, they are losing ground to Intel. You have said they have lost it already. Now we all now that you and I have different view on this. It really doesnt matter as the only reason I brought it up to begin with was to help illustrate reasons into Intel's and AMD's CPU roadmaps and why they are they way they are. That was fairly clear and you only refute it for reasons that only you know. Again, I could care less about it. The point was served.

My main point about consumers not getting 8 core system was based off very real logistics. It costs moneh to make things like this happen. Money is only spent on a business venture as an investment. That investment is expected to make a return and a profit. You seem to think that money is no object in the business world. That Intel and AMD only want to produce 8 core systems just because they can and the consumer market has a dire need for them. You live in a complete fantasy world. When it comes to computers, the consumer market is driven by the business world. End of story. There is nothing you can say that can change this, no matter how much you want to believe otherwise. Consumer grade 45nm based 8 cores CPUS are not going to be all the rage in less than a year. It's simply NOT going to happen. If you wish to continue say they will, then please do. I (and anybody living in reality) will simply wait until the time is right for them. Right now it isnt, and it simply wont be in the upcomming months. Years maybe, not months. I mean, we havent even seen consumer mobo's with dual CPU slots for quads yet and yet you seem to insist we will skip right over that and go straight to 8 cores on one chip and Intel and AMD will make a killing. No sir. It wont happen that way, and if it did it would be a first in the computer market. Hardware has always had to wait on software to push the hardware to the next level. Right now software is not keeping up with the hardware.

We have indeed hit a wall with clock speed. It is a financial wall and not one of science. There is simply no reason current CPU's could not be made to run at four or even five gigahertz. Infact some people have done that with consumer grade CPU's. The problem is nobody wants to buy, install, and run a bloody heat exchanger to do it. It's not a practical solution at all. But it can be done. My point hasnt changed. Again my point is this:

There simply is no real demand for a consumer grade 8 core CPU. There will not be a demand until these criteria are met.

1. Software makes a demand.

2. A practical price point can be met in the manufacture of these chips.

3. A practical pirce point can be met in the use of supporting hardware (ie mobo, cooling, PSU, etc).

Once all of these criteria are met then yes we will have 8 core CPU's. Until then, we will do what we have always done in the consumer based micro market. We will use 3rd party solutions. Things like mobo's with dual CPU sockets are among the most common solution. Look at the AM2 boards for a recent example. Given the fact that we REALLY have to look hard to justify quads in the consumer market I just dont see the 8 core demand taking off like a rocket. As I pointed out in my orginal post there is a 8 CPU option to us, but very few people seem to be rushing out ot buy two quad core opterons. Yes, they sell, but it's not like they any retailer like newegg is ever out of stock. Again, there is no real demand.

So inshort, you insist that Intel and AMD will rush to market a consumer based 8 core CPU in about a year. You give no reasons for this other than to say my pragmatic market views are unrelated and therefore wrong. Ok, you win. I'll join you in fantasy land to look at they pretty rose colored world and stand buy to get my new 8 core CPU in about one years time. Dont me angy with the fact I will have a bit of a nagging feeling we wont see them. Something tells me there is a vague chance that we may be waiting until a later date. Dont know why that is, as in fantasy world all our fondest wishes come true. ^_^

Do you even know what IGPs are?
Integrated graphics...

Go look at any sales numbers for GPUs and you will see that Intel has had the most marketshare for quite awhile. Some Q4 '07 sales numbers. So go ahead, try to make up more facts...

Yes, I obviously don't understand money or the business world... I don't work at a financial institution nor do I attend the ASU WP Carey School of Business. You are assuming too much and it shows. I have never said anything about octocores becoming mainstream in the next year, I have simply stated that they will be released in roughly a year.
You seem to either ignore, misunderstand or simply twist everything I post.

I, yet again, have to tell you that you need to research more before you make asinine statements. There are quite a few dual and even quad socket motherboards that will accept quadcores. So please do yourself a favor and stop acting like you know everything.

So if software is always the thing that drivers hardware? Why have we already moved from dualcores to quadcores while the vast majority of applications are still not multithreaded? Seems to be a flaw in your logic there...

Actually, NO, the wall is not due to finances, it is due simply to SCIENCE. Go read up on extreme overclocking, Phase, LN2, DICE, chilled water, TECs, they all can push current CPUs to ridiculous speeds and keep them cool but it takes a large amounts of volts. That high amount of voltage actually degrades the silicon and causes it slow down and react less to the extra voltage. One of the reasons you don't see extreme overclockers keeping their CPUs for too long. Plus there is also the fact that most people don't have the quality components or time to actually be able to run CPUs stable at those speeds even if they had the necessary cooling. You also need to remember that most of the time these are nearly suicide runs or quick benches for a few hours, these are not 24/7 clocks and voltages.

So to counter your points-
#1 Software doesn't always dictate hardware, my multithreading example.
#2 Practical costs have already occured, at least on Intel's side and things won't be changing when Nelahem arrives. I also expect AMD's quadcores to do just fine once they mature the 65nm process and eventually move to 45nm.
#3 When have motherboards, cooling, PSUs, with the exception of the extreme ones, been out of a normal price range? With the eventual die shrinks, we are talking about octocores with a TDP less than 180w, motherboards will be released/updated accordingly and cooling might need a small update for the highend octocores.

They are rushing nothing to the market, they are simply following their roadmaps. Just because you don't like what I have to say doesn't make it wrong...

all too long, who is going to bother to read these ramblings, this is a forun, not a PhD program. 2006toohuman