apology accepted. and I apologize if my tone offended you in the past Idontcare.
no it doesn't. using four dimms would increase the memory bandwith, but that is completely useless because the FSB would become the limiting factor. Also the latency of the modules would increase even more with 4 modules, and that is what really hurts the performance.
come on nemesis just admit it, skulltrail isn't the game/everyday application crusher you though it would be. hats off to encoding and 3d rendering performance though!
- Dec 30, 2006
- 11,366
- 2
- 0
Go toanother thread . So if you ran those other 1 socket systems with just 1 memory stick they would be faster ? and have less latency is that correct.
So single channel is better than dual channel and dual channel is better than q channel .
Feeding 1 quad core with 1 dim doesn't seem right some how. But your the expert.
So single channel is better than dual channel and dual channel is better than q channel .
Feeding 1 quad core with 1 dim doesn't seem right some how. But your the expert.
- Dec 30, 2006
- 11,366
- 2
- 0
My daughter and myself were discussing skulltrail last night. I was saying its a nice setup but really who needs it. I was a little surprized when she said she does.
She a pretty good hack. SO I said what on earth do you need this thing for.
She said remember how you said that opertunities in new software are a little more limited than they were 15 years ago. I replied ya. So .
She said SK offers her an opertunity at the first level of development that I said she would never have. I said OK ya so you can work on development of multi-thread apps. She LOL said ya that to . But she said your forgetting what Intel wants to do in gaming.
She said she needs SK to work On RTRT for gaming apps. I said ya for RTRT the more cores the better. Its true. Well she said ST offers game developers the opertunity to work with a high powered machine just for that purpose.
Than she said we need to get a sk system befor their all sold out . I said not to worry they won't sell out.
She said BS. I am not the only programmer out there that knows this. She said they will sell out right away. SO i will order her one. But will see if they sell out. She could be right,
We are looking at this system all wrong on the forums according to her. She said to work on Intels future gaming direction in programming you need tools today. She says this is the best tool out right now for work on RTRT that wont require ganging systems together.
SO i guess the real question here is this. Does SK offer programmers an opertunity to develop software at the 1st level of development? IN RTRT Gaming it most certainly does.
Its like she said . This SK platform isn't for gamers or for that matter Benchers.
She said this is the best Workstation ever created. As a programmer she sees opertunity with skull trail. I said you can get other systems a lot cheaper. She said true . But SK offers O/Cing so programmers can check out how new code scales with higher performance.
I don't know I am not that good with software so I haven't a clue. But she says it will make a differance.
So basicly is what she said is this. Get me a system now because these will sell out fast. Because others see this opertunity also.
Well find out shortly. If intel is only making 2,000 of these they should sell out right away. If she is correct. I know it will be useful for developing multi thread apps. But she may be right in that game developers are going to snap this systems up. Beings how it is the only system with intel chipset that has XF and sli. Game developers will buy these workstations to work on todays games as well as futete RTRT development.
She a pretty good hack. SO I said what on earth do you need this thing for.
She said remember how you said that opertunities in new software are a little more limited than they were 15 years ago. I replied ya. So .
She said SK offers her an opertunity at the first level of development that I said she would never have. I said OK ya so you can work on development of multi-thread apps. She LOL said ya that to . But she said your forgetting what Intel wants to do in gaming.
She said she needs SK to work On RTRT for gaming apps. I said ya for RTRT the more cores the better. Its true. Well she said ST offers game developers the opertunity to work with a high powered machine just for that purpose.
Than she said we need to get a sk system befor their all sold out . I said not to worry they won't sell out.
She said BS. I am not the only programmer out there that knows this. She said they will sell out right away. SO i will order her one. But will see if they sell out. She could be right,
We are looking at this system all wrong on the forums according to her. She said to work on Intels future gaming direction in programming you need tools today. She says this is the best tool out right now for work on RTRT that wont require ganging systems together.
SO i guess the real question here is this. Does SK offer programmers an opertunity to develop software at the 1st level of development? IN RTRT Gaming it most certainly does.
Its like she said . This SK platform isn't for gamers or for that matter Benchers.
She said this is the best Workstation ever created. As a programmer she sees opertunity with skull trail. I said you can get other systems a lot cheaper. She said true . But SK offers O/Cing so programmers can check out how new code scales with higher performance.
I don't know I am not that good with software so I haven't a clue. But she says it will make a differance.
So basicly is what she said is this. Get me a system now because these will sell out fast. Because others see this opertunity also.
Well find out shortly. If intel is only making 2,000 of these they should sell out right away. If she is correct. I know it will be useful for developing multi thread apps. But she may be right in that game developers are going to snap this systems up. Beings how it is the only system with intel chipset that has XF and sli. Game developers will buy these workstations to work on todays games as well as futete RTRT development.
No one is saying SK is worthless, merely saying whatever worth you intend to extract from it can be done by purchasing an alternative which is already available and for less money.
What does programming on a dual-socket 8-core Penryn board hamstrung by FBDIMM have to do with SLI or CF on a 4-core/8-thread DDR3 Nehalem next year?
The future of gaming is not going to be with Penryn or dual-socket or FBDIMM.
The point is simply this - for any argument you can make as to why SK is a good idea for programming can likewise be made for any standard off-the-shelf dual-socket workstation motherboard...only for a ton less coin invested into the mobo and QX processors.
I like the idea of SK, the concept was neat one year ago when there wasn't anything else going on. But fast forward a year, and be less than a year away from Nehalem release, and you got yourself a real cost/performance issue with SK.
It has all the downsides that the orginal QuadFX "Quad Father" had - too much power consumption, too little OC overhead, shackled to memory standard that enthusiasts have NEVER enthused over, and gaming performance that is single-digit improvement over alternatives that cost half as much.
What does programming on a dual-socket 8-core Penryn board hamstrung by FBDIMM have to do with SLI or CF on a 4-core/8-thread DDR3 Nehalem next year?
The future of gaming is not going to be with Penryn or dual-socket or FBDIMM.
The point is simply this - for any argument you can make as to why SK is a good idea for programming can likewise be made for any standard off-the-shelf dual-socket workstation motherboard...only for a ton less coin invested into the mobo and QX processors.
I like the idea of SK, the concept was neat one year ago when there wasn't anything else going on. But fast forward a year, and be less than a year away from Nehalem release, and you got yourself a real cost/performance issue with SK.
It has all the downsides that the orginal QuadFX "Quad Father" had - too much power consumption, too little OC overhead, shackled to memory standard that enthusiasts have NEVER enthused over, and gaming performance that is single-digit improvement over alternatives that cost half as much.
- Dec 30, 2006
- 11,366
- 2
- 0
Your a very good poster. But my daughter was right. We forum junkies aren't seeing this thing correctly.
For software development the FB dimms are perfect. Offer stability and loads of bandwidth. As I said I don't know if you need that kind of bandwidth. But as Dar. said.
People on the forums are viewing this worhstation in the wrong light. She read AT. review as I did.
She said that reviewer was very good and very responsiable in the way presented SK to the enthusist community.
BUT! She said the reviewer could have pointed to the fact that in so far as game development this is the best system available. And that working on development of RTRT for intel this is a very good tool. Now the reviewer is aware of this . But all are saying the RTRT is years away . Intel says BS.
There reviewer did say its a good system to use for developing multi threaded apps.
But hardly anyone other than Intel seems to be taking RtRT seriously.
I have one thing to say to this .
When your in the jungle you don't Fu-c with an 800 pound gorilla.
To work on tomorrows games you need tools today. hats what this system offers FBdimms and all .
On the FBdimms. Micron is suppose to have some really fast RB dimms coming out.
This will change the picture a little . But not much.
I agree with ya. I agree with Dar also. Thats why I am getting a Wolf. for myself.
I didn't need any thing more than that. For epensis I can wait for Nehalem . If I last that long.
Fact is I am even holding off on Wolf. Why? Waits for new stepping. so one can achieve 4.5ghz with 1.4 volts or less. I think most seen were these 2 socket systems are running at 4.ghz. at stock v. For some reason AT had to raise V a little but others had no problems at stock. I know what I want and am willing to wait for it. Besides I want X48 with wolf. Hell who knows if intel comes out with a go type stepping for wolf. Who knows were this chip. will stop scaling. All I can say is this.
I been reading all about the high voltages required to Overclock above 4.ghz.
For SK to O/C @4ghz at stock is incrediable. So unlike early adopters I wait a little while get the good stuff and come in the forums and beable to say . Whats your problem . I am setting at 4.5 ghz perfectly stable and cool and low V. Early adapters complain about this and that. Patients is a virue. Early adapters are going to have to except what they have or spend More money. Guys that have Conroe right now don't need penryn right now . Wait awhile. Get the good stuff. When Nehalem comes out do the same thing wait it out let the early adapters have the problems befor jumping in.
You will end up with a system that pawns theirs at lower cost. and skip all the problems.
For software development the FB dimms are perfect. Offer stability and loads of bandwidth. As I said I don't know if you need that kind of bandwidth. But as Dar. said.
People on the forums are viewing this worhstation in the wrong light. She read AT. review as I did.
She said that reviewer was very good and very responsiable in the way presented SK to the enthusist community.
BUT! She said the reviewer could have pointed to the fact that in so far as game development this is the best system available. And that working on development of RTRT for intel this is a very good tool. Now the reviewer is aware of this . But all are saying the RTRT is years away . Intel says BS.
There reviewer did say its a good system to use for developing multi threaded apps.
But hardly anyone other than Intel seems to be taking RtRT seriously.
I have one thing to say to this .
When your in the jungle you don't Fu-c with an 800 pound gorilla.
To work on tomorrows games you need tools today. hats what this system offers FBdimms and all .
On the FBdimms. Micron is suppose to have some really fast RB dimms coming out.
This will change the picture a little . But not much.
I agree with ya. I agree with Dar also. Thats why I am getting a Wolf. for myself.
I didn't need any thing more than that. For epensis I can wait for Nehalem . If I last that long.
Fact is I am even holding off on Wolf. Why? Waits for new stepping. so one can achieve 4.5ghz with 1.4 volts or less. I think most seen were these 2 socket systems are running at 4.ghz. at stock v. For some reason AT had to raise V a little but others had no problems at stock. I know what I want and am willing to wait for it. Besides I want X48 with wolf. Hell who knows if intel comes out with a go type stepping for wolf. Who knows were this chip. will stop scaling. All I can say is this.
I been reading all about the high voltages required to Overclock above 4.ghz.
For SK to O/C @4ghz at stock is incrediable. So unlike early adopters I wait a little while get the good stuff and come in the forums and beable to say . Whats your problem . I am setting at 4.5 ghz perfectly stable and cool and low V. Early adapters complain about this and that. Patients is a virue. Early adapters are going to have to except what they have or spend More money. Guys that have Conroe right now don't need penryn right now . Wait awhile. Get the good stuff. When Nehalem comes out do the same thing wait it out let the early adapters have the problems befor jumping in.
You will end up with a system that pawns theirs at lower cost. and skip all the problems.
- Dec 30, 2006
- 11,366
- 2
- 0
I would like to point out 1 game were SK showed up big time. Now if thats your game of choice. LOL! This is the system for you. $$$$$$$$$$$$. Lost Planet. Now I played with the demo version . Wasn't real impressed with porting to pc in keyboard function. But I am sure that was fixed or should have been , Did I like the game itself . You bet I did.
So what game comes out next that will change our perspective of what this system offers. Right NOW! LOST PLANET= ST.
So what game comes out next that will change our perspective of what this system offers. Right NOW! LOST PLANET= ST.
Nemesis these are all good arguments you are making but nonetheless they are the types of arguments you make when you are attempting to convince someone else to buy the computer for you.
No one makes these kinds of arguments to psyche themselves up to buy the computer for themselves. If you want 8-cores and FBDIMM to program with then you'd get a workstation board, a set of Xeon processors, and some decently affordable DDR2-667 FBDIMMs.
Since leaving TI my income is now derived from the fruits of my labor in computer programming. I create code that runs realtime in analyzing financial market data for speculative trading. I have 6 quad-core systems running 24x7 on the matter.
No way in hell if I am footing the bill could I ever justify or make an argument that I need 2 quad-cores (versus just 1 quad-core) in a single system before I could be effective in building multi-threaded apps...unless I was trying to convince my clients to foot the bill or if I was trying to convince my boss to authorize the expense thru a cost-center.
It just doesn't make sense. SK doesn't suddenly enable you to cross some magical threshold where you can effectively accomplish some task which was hitherto impossible.
But if the system can be effectively free because you are making arguments for someone else to foot the bill then sure who would turn down the performance at that point? The performance/cost ratio can become quite compelling in the asymptotic limit of the cost approaching zero.
I pity the person that has convinced themselves that having a Skulltrail system will enable them to accomplish something that they otherwise could have never accomplished with standard dual-socket FBDIMM workstation or standard i780&SLI or x48 & CF systems.
No one makes these kinds of arguments to psyche themselves up to buy the computer for themselves. If you want 8-cores and FBDIMM to program with then you'd get a workstation board, a set of Xeon processors, and some decently affordable DDR2-667 FBDIMMs.
Since leaving TI my income is now derived from the fruits of my labor in computer programming. I create code that runs realtime in analyzing financial market data for speculative trading. I have 6 quad-core systems running 24x7 on the matter.
No way in hell if I am footing the bill could I ever justify or make an argument that I need 2 quad-cores (versus just 1 quad-core) in a single system before I could be effective in building multi-threaded apps...unless I was trying to convince my clients to foot the bill or if I was trying to convince my boss to authorize the expense thru a cost-center.
It just doesn't make sense. SK doesn't suddenly enable you to cross some magical threshold where you can effectively accomplish some task which was hitherto impossible.
But if the system can be effectively free because you are making arguments for someone else to foot the bill then sure who would turn down the performance at that point? The performance/cost ratio can become quite compelling in the asymptotic limit of the cost approaching zero.
I pity the person that has convinced themselves that having a Skulltrail system will enable them to accomplish something that they otherwise could have never accomplished with standard dual-socket FBDIMM workstation or standard i780&SLI or x48 & CF systems.
- Dec 30, 2006
- 11,366
- 2
- 0
You make good points. But this time your being a little on fair to me. I don't recall having every gone into a thread that ask what should I buy this or that. My post history will show that.
I don't care what peole buy. I only care about the tech.
Here is an example . and its perfect for what you just stated.
http://www.techreport.com/articles.x/14052/6
Notice how well phenom does in one bench in lost planet. Than look at the one were SK stands out big time 10% that to me is big.
All am saying is we don't know what games will be out tomorrow. We don't know how long an 8 core system will offer future proofing. You don't have to have a Nehalem to take advantage of 8 cores when the time comes. I am planning on using Larrobee when every its available. With ST right now their will be room for 4 such larrobee gpu's as far as scaling if 16 cores make the systen scale . I am pretty confident that intel will be able to make 4 cards scale. That would 64 cores working on your gaming system . If intel can gett 1 game out that uses RTRT with the release of larrabee that is a great game . That will be enough to get people to buy these. So the future proof thing is unknown . So are many other factors. Basing an opinion on whats known today is being a little short sited . ST may have a lot longer life than you think. Maybe not. Thats all I am saying. I don't care who buys these things . never have never will. But I am not short sited either. I see RTRT as near future when most others see it as being along way off. I guess will learn more with what Future Mark is going todo with its gaming IP . But if I were them . I would look at the 800pound gorilla and go with what they want. Its Future marks opertunity to come into the gaming market and go right to the top . RTRT is the only way that will happen. We look at Intel and say ya they want RTRT. But I lok at INTEL /Apple and say they need an IP for RTRT. Someone is going to step up to the plate and make a shi-t load of money. LooK at Dec. wait you can't beause they no longer exist. Shorted sighted they were.
We don't know what uber highend Nehalem single socket is going to cost. But from what we do know. There are going to be 2 sockets for the Desktop Nehalem . But I suspect when we find out the forums are going to moun big time. Your not going to beable to have a top end gamer for the cheap. Guys are going to bitch bigtime about it. Hay buy what you can afford and be happy with it.
I like the way Nehalem is going . I hate to buy top of the line stuff. Only to have unstable low end stuff show good results at high O/Cs. Intel is doing it right . If I spend the money for top of the line . Your not going to be able to match performance with a lowend desktop. This is good not bad . But many will disagree with that.
You say I sound like I am selling the ST system . Wouldn't I be just as correct in stateing that your trying to sell Nehalem without kmowing what socket 1366 is going to cost. Your not going to get this processor for under a $1000. Socket 1066 yes it will be cheaper. But will it outperform ST? Socket 1366 I am sure will . Socket 1066 I don't know . If socket 715 is real . That won't compete at all.
As I said I only care about the tech and discussions on it. I don't care if I buy this tech or if anyone else does.
What I do care about is RTRT. St offers developers a tool to work with. that is O/C able so developers can easily check frequency scaling.
Maybe I am reading to much into it. But I would rather my glass be full and overflowing than empty.
I don't care what peole buy. I only care about the tech.
Here is an example . and its perfect for what you just stated.
http://www.techreport.com/articles.x/14052/6
Notice how well phenom does in one bench in lost planet. Than look at the one were SK stands out big time 10% that to me is big.
All am saying is we don't know what games will be out tomorrow. We don't know how long an 8 core system will offer future proofing. You don't have to have a Nehalem to take advantage of 8 cores when the time comes. I am planning on using Larrobee when every its available. With ST right now their will be room for 4 such larrobee gpu's as far as scaling if 16 cores make the systen scale . I am pretty confident that intel will be able to make 4 cards scale. That would 64 cores working on your gaming system . If intel can gett 1 game out that uses RTRT with the release of larrabee that is a great game . That will be enough to get people to buy these. So the future proof thing is unknown . So are many other factors. Basing an opinion on whats known today is being a little short sited . ST may have a lot longer life than you think. Maybe not. Thats all I am saying. I don't care who buys these things . never have never will. But I am not short sited either. I see RTRT as near future when most others see it as being along way off. I guess will learn more with what Future Mark is going todo with its gaming IP . But if I were them . I would look at the 800pound gorilla and go with what they want. Its Future marks opertunity to come into the gaming market and go right to the top . RTRT is the only way that will happen. We look at Intel and say ya they want RTRT. But I lok at INTEL /Apple and say they need an IP for RTRT. Someone is going to step up to the plate and make a shi-t load of money. LooK at Dec. wait you can't beause they no longer exist. Shorted sighted they were.
We don't know what uber highend Nehalem single socket is going to cost. But from what we do know. There are going to be 2 sockets for the Desktop Nehalem . But I suspect when we find out the forums are going to moun big time. Your not going to beable to have a top end gamer for the cheap. Guys are going to bitch bigtime about it. Hay buy what you can afford and be happy with it.
I like the way Nehalem is going . I hate to buy top of the line stuff. Only to have unstable low end stuff show good results at high O/Cs. Intel is doing it right . If I spend the money for top of the line . Your not going to be able to match performance with a lowend desktop. This is good not bad . But many will disagree with that.
You say I sound like I am selling the ST system . Wouldn't I be just as correct in stateing that your trying to sell Nehalem without kmowing what socket 1366 is going to cost. Your not going to get this processor for under a $1000. Socket 1066 yes it will be cheaper. But will it outperform ST? Socket 1366 I am sure will . Socket 1066 I don't know . If socket 715 is real . That won't compete at all.
As I said I only care about the tech and discussions on it. I don't care if I buy this tech or if anyone else does.
What I do care about is RTRT. St offers developers a tool to work with. that is O/C able so developers can easily check frequency scaling.
Maybe I am reading to much into it. But I would rather my glass be full and overflowing than empty.
- Dec 30, 2006
- 11,366
- 2
- 0
That was completely on fair guys. Don't ever confuse my enthisaim for tek and the 1 company I feel that will get us their the fastest. With GOD like worship. Its a cheap shot in the lowest form.
It made me LOL tho! Thanks for that. See how hard the people of the world LOL about GOD when the coastal cities are all under water. Than poke fun at GOD fearing people.
It made me LOL tho! Thanks for that. See how hard the people of the world LOL about GOD when the coastal cities are all under water. Than poke fun at GOD fearing people.
Ah you are mis-interpretting what I am saying. I do not mean to imply "you, nemesis" are trying to convince someone to buy the hardware for "you" or that "you" are trying to rationalize buying the hardware for yourself.
I am using the term "you" and "a person" etc in reference to some hypothetical person out there right now in the world who would be asking themselves "how do I convince someone that I am being rational about buying, or asking them to procure for me, a skulltrail system?".
My point is that the argument is done before it gets started. There is no rational for skulltrail other than pure emotional "I got to have it!".
Any logical reason you can concoct to justify needing the performance (be it for gaming, crunching, massive multi-threaded rendering, or test-bed for code development) can be remedied by purchasing an existing system already designed for the purpose in that market segment.
Skulltrail was designed to fight competition in a new market segment, one that never materialized because AMD killed the quad-father. Without a market segment to fit into the Skulltrail product is now going to try and fight for justification by elbowing its way into pre-existing market segments. The problem is that those pre-existing market segments already have their own top-of-the-line optimized uber products and skulltrail simply adds cost not performance to those market segments.
Just like Killer NIC.
- Dec 30, 2006
- 11,366
- 2
- 0
Ok ! I see what your saying. I really didn't know their were sytems available that developers could work with that had multipol PCI-E slots that RTRT developers could use without ganging . Please except my apolgy. So that makes ST utterly useless . I can live with that . So long as RTRT developers have systems that can do the same at less cost.
I also thought that for game developers a Intel chipset could be used to develop games using either NV sli or CF in that development. Beings how intel has 78% market share I thought maybe developers would want a system that could do both sli and Xf in the development stage. Now that you set me straight on that I guess your right.
I don't like the way crysis was developed at all . Made for NV. using NV chipsets and sli. It would have been better to use ATI XF with and AMD cpu and chipset for best performance. I just figured using intel chipset and Cpu and running either Sli or XF for development would be the best for the majotity rather than minority.
But I forgot we now live in a country that caters to the minorities and the hell with the majority.
I also thought that for game developers a Intel chipset could be used to develop games using either NV sli or CF in that development. Beings how intel has 78% market share I thought maybe developers would want a system that could do both sli and Xf in the development stage. Now that you set me straight on that I guess your right.
I don't like the way crysis was developed at all . Made for NV. using NV chipsets and sli. It would have been better to use ATI XF with and AMD cpu and chipset for best performance. I just figured using intel chipset and Cpu and running either Sli or XF for development would be the best for the majotity rather than minority.
But I forgot we now live in a country that caters to the minorities and the hell with the majority.
They are available but in the sense that you typically buy hardware to address a specific issue in the business environment, skulltrail allows you to buy hardware and apply it to any issue. It is the more generic platform, no question there.
Developers right now might have a system for CF or a system for SLI and yes you can do RTRT (realt-time raytracing) on them of course because the argument is merely one of scaling resolution and numbers of objective definitions, but they don't have a system that can be CF today and SLI tomorrow.
Skulltrail does afford that option, although I can't think of any compelling reason why a serious developer would burden themselves with having to manually swap out their video cards rather than just have more than one platform at their disposal.
Here's my concern with attempting to support any business utilization argument on the premise that skulltrail is the solution...namely the fact that it is simply not targeted at the profesional business environment.
You know what is on Nvidia's mind when they are selling Quadro's. The product is targeted at the professional business market. But Skulltrail?
Skulltrail isn't being released as the freelance graphics artists solution to beating a deadline or snagging that storyboard to pay the next month's rent. It isn't being released with the idea that the next Gears of War is going to be developed by a farm of coders each operating with their own skulltrail system tucked away under their desk in their cubicle.
Skulltrail is being released because it supposedly represents the end-all be-all of overclocking/gaming/enthusiast segment. As such it ought to offer compelling advantages and performance to today's enthusiast. Maybe if it was DDR3 then you could see it being sought after. I just don't see enthusiasts going after an FBDIMM board, for simple fact they never have.
Developers right now might have a system for CF or a system for SLI and yes you can do RTRT (realt-time raytracing) on them of course because the argument is merely one of scaling resolution and numbers of objective definitions, but they don't have a system that can be CF today and SLI tomorrow.
Skulltrail does afford that option, although I can't think of any compelling reason why a serious developer would burden themselves with having to manually swap out their video cards rather than just have more than one platform at their disposal.
Here's my concern with attempting to support any business utilization argument on the premise that skulltrail is the solution...namely the fact that it is simply not targeted at the profesional business environment.
You know what is on Nvidia's mind when they are selling Quadro's. The product is targeted at the professional business market. But Skulltrail?
Skulltrail isn't being released as the freelance graphics artists solution to beating a deadline or snagging that storyboard to pay the next month's rent. It isn't being released with the idea that the next Gears of War is going to be developed by a farm of coders each operating with their own skulltrail system tucked away under their desk in their cubicle.
Skulltrail is being released because it supposedly represents the end-all be-all of overclocking/gaming/enthusiast segment. As such it ought to offer compelling advantages and performance to today's enthusiast. Maybe if it was DDR3 then you could see it being sought after. I just don't see enthusiasts going after an FBDIMM board, for simple fact they never have.
- Dec 30, 2006
- 11,366
- 2
- 0
Again Idontcare you make all solid points. Now as for developers switching out cards . Ya your right . But that doesn't mean they cant use 2 setups. one with sli one with XF. So that the chipset and cpu are run on the best platform or system. Changing the GPU only rather than using NV chipset and AMD chipset. Just use intel chipset and you have it all
I didn't know there were alot of game developers interested in RtRT. Intel is interested in it because it scales well with mult- cores. This is a race man. Intel is trying to get to RTRT before NV gets Cuda. But Havik I believe was helping in cudas development. Intel now owns that. Along with that shut down . It killed ATI/NV physics hopes threw software. Weres as RTRT is great for collision detection which we need for more physics in game play.
Intel is playing cut throat with a hand containing same suit AKQ. The only thing going to beat that is same hand in hearts.
The next year is going to be great. Spring IDF is going to be great. As far as more Nehalem info.
But Fall IDF is going to be an eye opener when intel shows larrabee working with Nehalem or Penryn either one. Were going to here alot more about RTRT in the coming months. IF intel does Go RTRT road exspect that to be the topic of discussion in forums after spring IDF. ST as far as future proof all hinges on larrabee. If larrabee is good. people will get excited about an old ST system.
No way do I exspect Larrabee to = NV or ATI top end. I do however believe it will scale better with Multipol cards better than both ATI and NV. Right off the bate. Much like CSI is going to be better that AMD HT3 right off the bat.
I didn't know there were alot of game developers interested in RtRT. Intel is interested in it because it scales well with mult- cores. This is a race man. Intel is trying to get to RTRT before NV gets Cuda. But Havik I believe was helping in cudas development. Intel now owns that. Along with that shut down . It killed ATI/NV physics hopes threw software. Weres as RTRT is great for collision detection which we need for more physics in game play.
Intel is playing cut throat with a hand containing same suit AKQ. The only thing going to beat that is same hand in hearts.
The next year is going to be great. Spring IDF is going to be great. As far as more Nehalem info.
But Fall IDF is going to be an eye opener when intel shows larrabee working with Nehalem or Penryn either one. Were going to here alot more about RTRT in the coming months. IF intel does Go RTRT road exspect that to be the topic of discussion in forums after spring IDF. ST as far as future proof all hinges on larrabee. If larrabee is good. people will get excited about an old ST system.
No way do I exspect Larrabee to = NV or ATI top end. I do however believe it will scale better with Multipol cards better than both ATI and NV. Right off the bate. Much like CSI is going to be better that AMD HT3 right off the bat.
Nemesis as you've done a great job documenting and noting here on the forum, Intel has been steadily bringing in-house or developing in-house all the tools they will need to dominate NVidia and ATI/AMD.
From the elbrus compilers to Havok physics engine to IMC on Nehalem with hyperthreading to Larrabee and 16 cores. (silverthorne and Sandy bridge fit in there somewhere too)
If they fail to dominate Nvidia and AMD in all market segments in 5years time then it will simply be because they failed in some massive manner ala Prescot style. Defeat is their's to snatch from the jaws of victory.
The resources being brought to bear on creating new products just simply dwarfs the competition, even with flawless execution by the competition you can't win when your playing second-fiddle to someone spending $4 on R&D for every $1 you spend.
What I want to know is what Nvidia has cooking to crack the CPU market. We know a good deal of what AMD thinks it needs to do to stay competitive, and likewise from Intel...but Nvidia is staying mum and you know they aren't stupid.
From the elbrus compilers to Havok physics engine to IMC on Nehalem with hyperthreading to Larrabee and 16 cores. (silverthorne and Sandy bridge fit in there somewhere too)
If they fail to dominate Nvidia and AMD in all market segments in 5years time then it will simply be because they failed in some massive manner ala Prescot style. Defeat is their's to snatch from the jaws of victory.
The resources being brought to bear on creating new products just simply dwarfs the competition, even with flawless execution by the competition you can't win when your playing second-fiddle to someone spending $4 on R&D for every $1 you spend.
What I want to know is what Nvidia has cooking to crack the CPU market. We know a good deal of what AMD thinks it needs to do to stay competitive, and likewise from Intel...but Nvidia is staying mum and you know they aren't stupid.
- Sep 28, 2005
- 21,087
- 3,596
- 126
nemisis, your daughter sounds awefully cute, however you need to tell her this.
go honey, have you worked with a server platform? do you understand what it means to have a dual cpu system?
Do you know the complete system will cost anywhere near the amount of 8000 dollars?
Watch her go okey daddy buy me a BMW instead.
and no im not insulting your daughter, i really think she's smarter then most girls ive talked to about computers.
exceptions would be rubycon. im sorry, but if she didnt post a pic, id swear she was a guy messing with us with multipul phd's in engineering.
But neminisis skulltrail is not for a programer, its for the hardcore elite gamer. If your daughter wants a multi core platform, the i5400 seaburg chipsets with dual E5430's might be a cheaper and better solution.
- Dec 30, 2006
- 11,366
- 2
- 0
Exactly right. Even tho I have heard Talk from Intel that Nehalem will be good. I am not positive on that at all. Beings how its the same team that developed Prescott. I won't hold my breath on that one at all. I remember Intels news releases on prescott and how far away from the truth those press releases were.
I am taking it on faith only . But I wouldn't wager on it. Intel has pretty much said it won't be much better than Penryn at same clocks. Core to core.
Intel can even screw Nehalem up and come out great if Larrabee is Good. Right now Larrabee is the big deal for intel to achieve its goals. Intel could limp threw another CPU screw up until . THE Israeli team gets Gesher done. These guys I believe in 100%.
The problem with NV is this . They will never ever get a X86 licese ever. They may have had a chance once . But because of SLI it will never happen . NV has to play nice.
ATI was smart. and did it correctly. I am sure AMD thought about shouting Intel out of the XF market. But I am pretty sure they found out that couldn't because of previsious deal with Intel . AMD backing ATI wasn't a deal breaker. Pretty sure Intel shared ring bus teck they have developed with ATI.
Its very much like when intel bought Elbrus. There was some concern that the tech sharing agreement that was inplace between Sun and Elbrus befor intel bought Elbrus could be Broken . But as it turned out Intel already had an agreement cross license with Sun . So it didn't matter.
I am taking it on faith only . But I wouldn't wager on it. Intel has pretty much said it won't be much better than Penryn at same clocks. Core to core.
Intel can even screw Nehalem up and come out great if Larrabee is Good. Right now Larrabee is the big deal for intel to achieve its goals. Intel could limp threw another CPU screw up until . THE Israeli team gets Gesher done. These guys I believe in 100%.
The problem with NV is this . They will never ever get a X86 licese ever. They may have had a chance once . But because of SLI it will never happen . NV has to play nice.
ATI was smart. and did it correctly. I am sure AMD thought about shouting Intel out of the XF market. But I am pretty sure they found out that couldn't because of previsious deal with Intel . AMD backing ATI wasn't a deal breaker. Pretty sure Intel shared ring bus teck they have developed with ATI.
Its very much like when intel bought Elbrus. There was some concern that the tech sharing agreement that was inplace between Sun and Elbrus befor intel bought Elbrus could be Broken . But as it turned out Intel already had an agreement cross license with Sun . So it didn't matter.
- Dec 30, 2006
- 11,366
- 2
- 0
Not to beat a dead horse but AMD should have followed Intels lead and developed X86 gpu in house. Rather than buy ATI.
If larrabee is good. The ATI buyout was a hugh mistake. Amd could have developed there own X86 gpu and stayed in the game.
The ATI deal could be the back breaker or it could be the step up AMD needed.
IT really comes down to Larrabee. If its GOOD AMD and NV are screwed if it sucks. Intel is going to feel the pain.
I believe ATI is using EPIC type code which is good for ATI. I haven't a clue what NV is using . But If X86 works out good in the GPU market. Its bad news for AMD and NV. and it could Kill both companies.
Why do I think ATI using EPIC code is good. Because thats exactly what I believe Larrabee will use with X86 decoder.
That would leave AMD with having develop a decoder and there still in the GAME.
If Intel uses X86 without an epic decoder. Than AMD is screwed.
It really all falls on a DAM GPU. Isn't that something. Amd new this that why Hector did what he did. Was it good smart. NO one can say until this hand has played out.
If larrabee is good. The ATI buyout was a hugh mistake. Amd could have developed there own X86 gpu and stayed in the game.
The ATI deal could be the back breaker or it could be the step up AMD needed.
IT really comes down to Larrabee. If its GOOD AMD and NV are screwed if it sucks. Intel is going to feel the pain.
I believe ATI is using EPIC type code which is good for ATI. I haven't a clue what NV is using . But If X86 works out good in the GPU market. Its bad news for AMD and NV. and it could Kill both companies.
Why do I think ATI using EPIC code is good. Because thats exactly what I believe Larrabee will use with X86 decoder.
That would leave AMD with having develop a decoder and there still in the GAME.
If Intel uses X86 without an epic decoder. Than AMD is screwed.
It really all falls on a DAM GPU. Isn't that something. Amd new this that why Hector did what he did. Was it good smart. NO one can say until this hand has played out.
The biggest concern I have with Nvidia is they are fabless, meaning they depend on the foundries UMC and TSMC to develop leading edge process technology.
This was not an achiles heel when their competitor was ATI, who was also fabless and thus could not gain a process technology advantage over Nvidia as they both used the same foundries.
ATI going with AMD gives them access to leading edge high performance logic fabs. (neither TSMC nor UMC are what you should consider to be high performance logic fabs)
So what is Nvidia going to do about this gap in their access to Intel caliber or even AMD caliber 45nm and 32nm tech nodes? Even if they design the next uber CPU (or CUDA) it will be produced on an inherently inferior technology node.
This was not an achiles heel when their competitor was ATI, who was also fabless and thus could not gain a process technology advantage over Nvidia as they both used the same foundries.
ATI going with AMD gives them access to leading edge high performance logic fabs. (neither TSMC nor UMC are what you should consider to be high performance logic fabs)
So what is Nvidia going to do about this gap in their access to Intel caliber or even AMD caliber 45nm and 32nm tech nodes? Even if they design the next uber CPU (or CUDA) it will be produced on an inherently inferior technology node.
- Dec 30, 2006
- 11,366
- 2
- 0
Idontcare this is the type of game I was referring to. It looks like it will require alot of Cpu and GPU power.
http://www.youtube.com/watch?v=4mBcS8vQL2Y
http://my.mashable.com/video/e...fedec485bf7ec09fcbccd5
There is a lot more stuff on this game at Utube it is a must see for us guys.
http://www.youtube.com/watch?v=4mBcS8vQL2Y
http://my.mashable.com/video/e...fedec485bf7ec09fcbccd5
There is a lot more stuff on this game at Utube it is a must see for us guys.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter