4x4 Quadfather = quad core MCP x 2!!!

[vitaminc]

http://www.theinquirer.net/default.aspx?article=35333

AMD follows Intel with Quadfather One

The die is cast

By Theo Valich: Thursday 26 October 2006, 08:28
WE MANAGED TO learn the official NDA expiry date and the launch of AMD's quad core offering. Or, should we say, its dual die dual socket chip.

After AMD laughed off Intel's dual core and upcoming quad core microprocessors as "not true multicore", with the main reason they physically share two dies, it seems to have done the same thing. Intel's Pentium D "Presler" consisted of two Cedar Mill cores, Core 2 Extreme "Kentsfield" features two Conroe cores and shared the front side bus.

Quadfather or the 4x4 as AMD marketeers have it, is actually a dual die quad core chip, using the Opteron processor socket and will launch on November 14th. The nForce 680a, the launch chipset is proof that the AMD-Nvidia strategic alliance still has four legs.

The launch line up was revealed here, but the pricing has been "re-constructed" severely downwards, with the fastest CPU pack now matching the price of Intel's Kentsfield. µ

Now Quadfather 4x4 should outperform Kentsfield :o 8 cores vs 4 cores!!!

[perkam]

http://www.theinquirer.net/default.aspx?article=35333


Now Quadfather 4x4 should outperform Kentsfield :o 8 cores vs 4 cores!!!8 cores ??? It's 4 cores vs 4 cores...it must be the trashed deerhound server proc that AMD scrapped to make way for K8L. (Basically two Rev F/Rev G cores on one die)

That's going to be some crappy proc...

Perkam

[gillll]

"with the fastest CPU pack now matching the price of Intel's Kentsfield. µ "


i do hope !

please don't let me buy an intel!!!

[cky2k6]

if this is true, octa core for less than 1 kentsfield!

[Syn.]

it dosnt make sence. QuadFather is when they put 2 duel core prorecessors on the same motherboard which has 2 socketts. INQ is saying that its actually now two chips put on the same die. I am sorry but wtf?

After INQ said that DX9L was going to be released on XP i dont trust a single word they say. Can we please introduce a rule that all INQ stories are named "Rumor"?

[gillll]

yup no one give another 4 cores for free.

[Cooper]

You forgot that AMD soon will reveal quad-core SocketF CPUs which will work in any "quadfather" mobo with just a BIOS flash

[perkam]

meh...all this excitement over AMD finding more ways to keep demand for Rev F/RevG from dropping like a nail in water ahead of K8L production.

Most likely Quadfather will be done smithfield style with two dual core 90nm dies on one proc...as OEMs will eat up the little 65nm shipments AMD does send out by q4 2006/q1 2007.

Perkam

[vitaminc]

8 cores ??? It's 4 cores vs 4 cores...it must be the trashed deerhound server proc that AMD scrapped to make way for K8L. (Basically two Rev F/Rev G cores on one die)

That's going to be some crappy proc...

Perkam

both are quad core MCP, but 4x4 platform is for dual sockets. :p

Kents will still dominate single quad core MCP vs. MCP k8, but 4x4 sounds like 2 K8 QC MCP for the same price as a QX Extreme kents.

[gillll]

maybe this is the first step of mp where the cpu is hybrid cpu and gpu ?

the matrix of load balancing possabilities in this way is much larger then current systems

u'll have cores on 2 sockets and no "vga card"
all of the cores will load balance the system
whereas today the system has battleneck due to the cpu/gpu distance.

[onewingedangel]

All the inq were saying is that after trashing intel for having to scrape together a product, amd did exactly the same for 4x4. Its still 2 sockets each with a dual core processor. Intel went with dual die single package MCM to slap together a non native quad core system for enthusiasts, and amd went with dual die dual package, dual socket.

The inq were either deliberately ambiguous with the language to generate traffic, or didn't properly check their work.

[nn_step]

All the inq were saying is that after trashing intel for having to scrape together a product, amd did exactly the same for 4x4. Its still 2 sockets each with a dual core processor. Intel went with dual die single package MCM to slap together a non native quad core system for enthusiasts, and amd went with dual die dual package, dual socket.

The inq were either deliberately ambiguous with the language to generate traffic, or didn't properly check their work.
Basically AMD slammed Intel for MCM
and the INQ is Slamming AMD for 2P

[uOpt]

Hm...

I think it is not very likely to be true (that it is dual socket quad-core = 8 cores).

But I wouldn't be too surprised either. I remember just recently betting with somebody that AMD would do the dual-die trick, too. It just makes too much sense if you have heat under control - the simplifications for the motherboard are substancial. Of course I made that bet when I said they do 8-core chips by combining two 4-core dies but who knows - maybe they use QuadFather as a testbed.

If they give me an overclockable 8-cores I'd buy it no matter what.

[freecableguy]

*Ahem.* Hate to break up this party but I must...

AMD CPUs communicates with the SPP via an HyperTransport tunnel, same as the SPP communicates with the MCP (via an HT tunnel). Placing 2 (dual-core or otherwise) dies on a single package without changing the number of CPU package pins means more than just 4 cores in one package. It also means that each die can only receive 1/2 of the maximum BW available from the PCI-e buses attached to the SPP or MCP. Same number of pins, same bus speed, same data availability....one HT link.

If you thought Intel's "antiquated" FSB design was BW limited wait until you see this. Intel knew what they were doing. They knew all they had to do was scale the FSB to get more BW. AMD has to "overclock" the IMC (or design a new one) in order to increase BW. As you can see, each design has its pros and cons. AMD is about to experience the con. ;)

Plus don't forget, the second CPU is going to be attached via an HT link to the first CPU (the SPP only allows for one HT link via the only set of pins available). This means that all the data throughput from the second CPU (and it's associated memory) as well as any video data must pass through the first CPU on their way to the PCI/PCI-e bus. *choke* ;)

-FCG

[nn_step]

*Ahem.* Hate to break up this party but I must...

AMD CPUs communicates with the SPP via an HyperTransport tunnel, same as the SPP communicates with the MCP (via an HT tunnel). Placing 2 (dual-core or otherwise) dies on a single package without changing the number of CPU package pins means more than just 4 cores in one package. It also means that each die can only receive 1/2 of the maximum BW available from the PCI-e buses attached to the SPP or MCP. Same number of pins, same bus speed, same data availability....one HT link.

If you thought Intel's "antiquated" FSB design was BW limited wait until you see this. Intel knew what they were doing. They knew all they had to do was scale the FSB to get more BW. AMD has to "overclock" the IMC (or design a new one) in order to increase BW. As you can see, each design has its pros and cons. AMD is about to experience the con. ;)

Plus don't forget, the second CPU is going to be attached via an HT link to the first CPU (the SPP only allows for one HT link via the only set of pins available). This means that all the data throughput from the second CPU (and it's associated memory) as well as any video data must pass through the first CPU on their way to the PCI/PCI-e bus. *choke* ;)

-FCG
Each Proc would have half the bandwidth.
Which with DDR2-800
is about as much as dual channel DDR2-400 for each. Which honestly isn't horrible but it doesn't help much

[iddqd]

*Ahem.* Hate to break up this party but I must...

AMD CPUs communicates with the SPP via an HyperTransport tunnel, same as the SPP communicates with the MCP (via an HT tunnel). Placing 2 (dual-core or otherwise) dies on a single package without changing the number of CPU package pins means more than just 4 cores in one package. It also means that each die can only receive 1/2 of the maximum BW available from the PCI-e buses attached to the SPP or MCP. Same number of pins, same bus speed, same data availability....one HT link.

If you thought Intel's "antiquated" FSB design was BW limited wait until you see this. Intel knew what they were doing. They knew all they had to do was scale the FSB to get more BW. AMD has to "overclock" the IMC (or design a new one) in order to increase BW. As you can see, each design has its pros and cons. AMD is about to experience the con. ;)

Plus don't forget, the second CPU is going to be attached via an HT link to the first CPU (the SPP only allows for one HT link via the only set of pins available). This means that all the data throughput from the second CPU (and it's associated memory) as well as any video data must pass through the first CPU on their way to the PCI/PCI-e bus. *choke* ;)

-FCG
Isn't overclock and scale essentially the same thing? Intel has to overclock their FSB, AMD has to scale their HTT bus. Meh.

(It is scalable!)

[perkam]

:confused: :confused: :confused: :confused: :confused:

Welcome to XS....

http://thinkmedia.blogs.com/think_media/images/gibberish.jpg

Perkam

[arisythila]

Lol

[3NZ0]

:confused: :confused: :confused: :confused: :confused:

Welcome to XS....

http://thinkmedia.blogs.com/think_media/images/gibberish.jpg

Perkam

Actually laughed out loud :D

[grimREEFER]

2 cloverton's is better

[Gag3]

2 cloverton's is better


:confused: :confused: :confused: :confused: :confused:

Welcome to XS....

http://thinkmedia.blogs.com/think_media/images/gibberish.jpg

Perkam

:rolleyes:

Intel Fanboy Alert.

[perkam]

:rolleyes:

Intel Fanboy Alert.I don't know if you can say that about grimreefer...I don't think anyone with a smithfield in their sig can possibly be an Intel fanboy :lol:

Perkam

[Charles Wirth]

Plus don't forget, the second CPU is going to be attached via an HT link to the first CPU (the SPP only allows for one HT link via the only set of pins available). This means that all the data throughput from the second CPU (and it's associated memory) as well as any video data must pass through the first CPU on their way to the PCI/PCI-e bus. *choke*

Very true, FCG please show a block diagram ;)

[nn_step]

Here is how I see what AMD would do, should they Choose MCM
http://img.photobucket.com/albums/v643/nn_step2/explaining.png

[freecableguy]

Very true, FCG please show a block diagram ;)

http://members.cox.net/kjboughton/FX_block.JPG

[ozzimark]

*Ahem.* Hate to break up this party but I must...
more than willing to see you try :toast:

AMD CPUs communicates with the SPP via an HyperTransport tunnel, same as the SPP communicates with the MCP (via an HT tunnel).
ok, nothing new yet :D

Placing 2 (dual-core or otherwise) dies on a single package without changing the number of CPU package pins means more than just 4 cores in one package.
i agree. this is what intel is doing. amd opted for the dual socket plan, for now.

It also means that each die can only receive 1/2 of the maximum BW available from the PCI-e buses attached to the SPP or MCP. Same number of pins, same bus speed, same data availability....one HT link.
oh no! dropped from 8gb/s to 4gb/s even though your analogy isn't entirely accurate. it's more like the bandwidth will be the same, but latency will be reduced if one cpu asks for something while the other cpu is already getting something. the block diagram of a dual socketF platform
http://www.amd.com/us-en/assets/content_type/DownloadableAssets/2P_S_WS_Comparison_PID_41460.pdf

see how there's one link connecting the two cpu's, then one link for each cpu to the chipset? ;)

If you thought Intel's "antiquated" FSB design was BW limited wait until you see this. Intel knew what they were doing. They knew all they had to do was scale the FSB to get more BW. AMD has to "overclock" the IMC (or design a new one) in order to increase BW. As you can see, each design has its pros and cons. AMD is about to experience the con. ;)
what's this about the memory controller? :confused: :confused:
even if you meant ht link, why do you think amd is simply moving to HT3.0, which offers.. ah, i forget. a lot more bandwidth :lol2:

Plus don't forget, the second CPU is going to be attached via an HT link to the first CPU (the SPP only allows for one HT link via the only set of pins available). This means that all the data throughput from the second CPU (and it's associated memory) as well as any video data must pass through the first CPU on their way to the PCI/PCI-e bus. *choke* ;)
already disproved this point, see link to amd's diagram :toast:
(which is more accurate than yours afaik, because to meet the "4x4" standard, quad sli must be supported, which means the 680a is used.. basically two 590's on one board... tehehe :D)

try harder next time you wanna crash my party, but you're more than welcome to party with us! :woot:

[Turtle 1]

Nice work freecable guy. You seem to have a lot stored on your PC/

I seen a diagram of how csi worked once. I was in a hurry and didn't bookmark it . I been looking for 3 days can't find . You wouldn't have it bookmarked would ya? But Intel may have been slow to the post but they have the winner. Even tho Tigerton isn't a complete CSI set up . It also has direct connect from the cpu to the nb.

[vitaminc]

Here is how I see what AMD would do, should they Choose MCM
http://img.photobucket.com/albums/v643/nn_step2/explaining.png

Quadfather is based on 90nm parts and AMD doesn't have HTT 3.0 on their 90nm.

They probably won't have HTT3.0 on 65nm, but they haven't disclosed a lot yet.

[informal]

Just to clear up a few things:

1)4x4 is not 2x Quad Core,it's 2 x Dual Core(until New Core launhes).

2)There is NO MCM for AMD and probabaly wont be until New Core launches(and the chances are there will never be a MCM from AMD,that is based on two identical cores)

3) There will be no RevF or RevG quar core of any sort(except the 4x4 Dual CPU platform)

So,having said that,I believe (and I am pretty sure) that AMD's 4x4 solution will bring forth a LOT of attention of enthusiasts and be among the fastest quad core solutions on the market(especially for gamers and those who do a lot of multitasking)

Good night to all

[Turtle 1]

Ozzimark I understand what your saying but I have a question . I don't know the ans. (in my case this is normal). Heres a qoute from your post.

-----------------------------------------------------------------------------------------------------------------------------------------
Freecable guy
Plus don't forget, the second CPU is going to be attached via an HT link to the first CPU (the SPP only allows for one HT link via the only set of pins available). This means that all the data throughput from the second CPU (and it's associated memory) as well as any video data must pass through the first CPU on their way to the PCI/PCI-e bus. *choke*
-----------------------------------------------------------------------------------------------------------------------------------------
Ozzimark
already disproved this point, see link to amd's diagram
(which is more accurate than yours afaik, because to meet the "4x4" standard, quad sli must be supported, which means the 680a is used.. basically two 590's on one board... tehehe )

try harder next time you wanna crash my party, but you're more than welcome to party with us

************************************************** ************************************************** ******
Even tho what you said is true I am confused. Why does Sli or even quad sli have those connectors between the cards. I always thought that even tho both cards or even 4 cards shared the work . That only one card sent all the info to the cpu . So it would be as freecable guy laid out as far as the rendered scene would be sent back to the cpu threw only1 16 pci-e lane and not both. Also I thought the nv solution for 2x16 lanes required both the Nb and SB . only ATI has 48 lanes on 1 chip. Or has nv released a new chipset with all pci-e lanes on 1 chip?? So the 4x4 is going to have 2 complete 590 chipsets on it. This is going to be one expensive complicated board.

Now if you were using 4 seperate card not sli. than I see your point . But I cant play 4 games at once. on 4 seperate lcd's.

Please explain or give link telling me this is incorrect thinking.

[illmatik]

My question in all of this is, will quadfather support AMD's Virtualization hypervisor aka Pacifica?

[Turtle 1]

Here's whats going to happen Intel is going to look at this 4x4 M/B and say yep . We can build a Tigerton type 2 socket setup . Within reasonable cost and use DDR3 memory and Give amd a heart attack.Funny thing is Intel has until K8L to change the NB memory controller. Kentsfield will do fine until AMD comes with native 4 core. Until than Intel has all the time they need.Also remember that Intel is bring PCI-e2 out in 07. Asus already has a 4 slot pci-e M/B out 8x8x4x8x Well have to wait and see if AMD can make hay with 4x4 or can intel turn this market niche into Candyland.

[gillll]

from what i see till now the world is finally being devided is systems designs.

amd is making quite a move larger then the introducing of the a64 cores. it's not just core design now but whole system.

i do hope amd will manage to do it because she hadn't done this before. and intel has the upper hand in introducing "news" but has also the lower one (btx?)
intel is larger after all.

/me holds hands for AMD

[uOpt]

Regarding splitting the 16 bit Hypertransport in two:

I recently discovered my work workstation ran at 8 bit. I set to 16.

Doesn't make a bloody difference. For a two-socket the overkill of the current hypertransport is so big it's hard to even touch it.

[ozzimark]

Even tho what you said is true I am confused. Why does Sli or even quad sli have those connectors between the cards. I always thought that even tho both cards or even 4 cards shared the work . That only one card sent all the info to the cpu . So it would be as freecable guy laid out as far as the rendered scene would be sent back to the cpu threw only1 16 pci-e lane and not both. Also I thought the nv solution for 2x16 lanes required both the Nb and SB . only ATI has 48 lanes on 1 chip. Or has nv released a new chipset with all pci-e lanes on 1 chip?? So the 4x4 is going to have 2 complete 590 chipsets on it. This is going to be one expensive complicated board.

Now if you were using 4 seperate card not sli. than I see your point . But I cant play 4 games at once. on 4 seperate lcd's.

Please explain or give link telling me this is incorrect thinking.
hm, that is a good point, i'm not entirely sure how video data is sent around with sli and two chipsets... i would PRESUME that amd/nvidia are smart enough to take advantage of both links, if possible.. but you're right, it might not simply due to the nature of how sli works. thanks for pointing that out, i'll look into it a bit closer. of course, FCG could always comment on it, since i know he knows a lot about sli/cf :toast:

Turtle 1, again with this memory controller talk. mind explaining in a bit further detail what you mean: "Funny thing is Intel has until K8L to change the NB memory controller."
what would they change about it?

[gillll]

http://www.techreport.com/onearticle.x/11099

http://techreport.com/etc/2006q4/tigerton/system3.jpg

in this aray of cpus with multi cores

Caneland server platform and Tigerton quad-core processor

does the system is like the 2p with one mem controller ?

or like the 4X4 for every socket mem controller ?

the pics support the one mem because of the large heast sink which probably is the mem controller.

then the 4x4 is really diffrenet design and as i had read here with ht3 will be better.

[Turtle 1]

Well in the caneland system FBD are used . I really would want DDR3 on a Intel 2 socket setup. So Intel would have to make the NB memory controller work with DDR3 for the desktop. They have the Chipset so really all they need to do is design a M/B for desltop use. But the power of these systems is unreal. This isn't full CSI but its Kinda like what Nehalem is bring to the table. But Nehalem will knock your socks off with your shoes on.

The Thing is Will Intel want to introduce this as the Top performance gaming rig. I think their pissed off enough at AMD that they will do it . I hope so anyway.

[Theli]

How high would the TDP of an 8-core 4x4 system be?
I've read 250W for the 4-core, which is pretty high.

[Order]

It will be 65nm, though, right?
That will lower the TDP.

[Turtle 1]

http://www.techreport.com/onearticle.x/11099

http://techreport.com/etc/2006q4/tigerton/system3.jpg

in this aray of cpus with multi cores

Caneland server platform and Tigerton quad-core processor

does the system is like the 2p with one mem controller ?

or like the 4X4 for every socket mem controller ?

the pics support the one mem because of the large heast sink which probably is the mem controller.

then the 4x4 is really diffrenet design and as i had read here with ht3 will be better.
Well I looked at the 2p system they said they had pics of but gues what . I counted 4 cpu's

There is only 1 memory controller. But 4 independent FSB . Along with that there are 4 point to point links between the cpu's and the NB 1 for each cpu.

I seen a diagram of this set up not to long ago. and the point to point connection at the Cpu was referred to as a node. 1 per cpu. It also said that if an 8 cpu system was used than a second memory controller would be used.

Now that confused me a bit as it said nothing about a point to point connection between the 2 memory controllers. Now info was scant at best but in a token ring set up which is what CSI is there would have to be a point to point connection between the 2 memory controller. Like I said this isn't true CSI but its along the same lines.

Any talk of HT3.0 being better than CSI is pure AMD fanboy BS. Anytalk of CSI being better than HT3.0 is pure Intel fanboy BS.

Since we can't test them against each other its unknown . I can say the CSI has nore bandwidth that HT3.0 but whats best we won't know tell Nehalem slaps K10 around like a DumB :banana::banana::banana::banana::banana:. Just like Yorksfield will do K8L

[Turtle 1]

How high would the TDP of an 8-core 4x4 system be?
I've read 250W for the 4-core, which is pretty high. I bad I see you said for 4 cores . I think amd can do that maybe lower

If AMD comes out with a 8 core 4x4 system rated at 250w Intel better re-engineer their system because that would impress me.

Here is all I can tell ya
X6800 = idle 154 load 202 Q6600 = idle 198 load 223 measured at the wall socket



So 8 cores for the tiger be about 475 at the wall socket. That figuring FBD . Also by tigers release Intel suppose to have
power consumption reduced.

[uOpt]

My 8-core 2.2 GHz socket 939 system takes 453 Watts max when doing octa-mpriming. Out of the A/C (with a OCZ GXS PSU), so you have to subtract the PSU efficiency.

Of course that's before overvolting.

Keep in mind the TDP has nothing to do with actual power consumption.

[nn_step]

If AMD comes out with a 8 core 4x4 system rated at 250w Intel better re-engineer their system because that would impress me.
How many times do I have to remind you Intel does nothing for the customers. They do things only because it makes their pockets fatter

[ozzimark]

Any talk of HT3.0 being better than CSI is pure AMD fanboy BS. Anytalk of CSI being better than HT3.0 is pure Intel fanboy BS.

Since we can't test them against each other its unknown . I can say the CSI has nore bandwidth that HT3.0 but whats best we won't know tell Nehalem slaps K10 around like a DumB :banana::banana::banana::banana::banana:. Just like Yorksfield will do K8L
the irony of this kills me :lol2: :lol2: :lol2: :lol2:

so what was that you were saying about fanboy BS again? you tested the parts already, thus why you can invalidate the statement you made just 2 sentences earlier? :D

edit: can you tell us the performance then? ;)

[biohead]

indeed ozzimark, this is one of those 'talk to the hand' moments.

[informal]

Any talk of HT3.0 being better than CSI is pure AMD fanboy BS.Anytalk of CSI being better than HT3.0 is pure Intel fanboy BS.
Since we can't test them against each other its unknown . I can say the CSI has nore bandwidth that HT3.0 but whats best we won't know tell Nehalem slaps K10 around like a DumB :banana::banana::banana::banana::banana:. Just like Yorksfield will do K8L

Yet another dumb quote/comment made just for some one's sig.Get it while it's hot.

[Turtle 1]

the irony of this kills me :lol2: :lol2: :lol2: :lol2:

so what was that you were saying about fanboy BS again? you tested the parts already, thus why you can invalidate the statement you made just 2 sentences earlier? :D

edit: can you tell us the performance then? ;)

I think I have confessed to being a Intel fanboy. I know Iam but what are you. Look fact is Intel slipped and fell on netburst we all know it, But the things they learned alon the way is going to pay off huge. Not that the billions they earned when falling killed them. Intel will never make the mistake again. K8l might maybe be a little better than conroe we don't know. AMD might regain the lead until yorkfield is released after that AMD will never regain the lead. K8L will be AMDs last shot as performance king. The Giant has awoken and shall never doze off again . IBM is who Intel has to watch. Maybe NV if they can survive.

[informal]

I think I have confessed to being a Intel fanboy. I know Iam but what are you. Look fact is Intel slipped and fell on netburst we all know it, But the things they learned alon the way is going to pay off huge. Not that the billions they earned when falling killed them. Intel will never make the mistake again. K8l might maybe be a little better than conroe we don't know. AMD might regain the lead until yorkfield is released after that AMD will never regain the lead. K8L will be AMDs last shot as performance king. The Giant has awoken and shall never doze off again . IBM is who Intel has to watch. Maybe NV if they can survive.

Oh look,another comment made for some one's sig.You have outdone yourself turtle boy,at least for today.Tomorrow is a new day,you can do even better :fact:

[vitaminc]

How many times do I have to remind you Intel does nothing for the customers. They do things only because it makes their pockets fatter

AMD isn't running an charity business either.

[nn_step]

AMD isn't running an charity business either.
exactly, favor which ever does what is important to you, the best.

[Turtle 1]

Ever since Conroe came out it been this.

1) spring IDF . Intel rigged the results.

2) Its really not fair to test graphics at low res because its not real world.

3) Its not that big of a lead only 20 to 30%

4) wait till K8l comes out it beat conroe by 40%

5) Torrenza will be the all everthing. Now even AMD is backing intel IBM solution . on the PCI-E

6) Csi doesn't have the bandwidth that direct connect HT3.0 has.

7) AMD still owns the server market . The highend is Intels Itanic the DP is Intels and with Tiger the server market will be intel.

I could go on and on . I could put tons of BS in my sig. But I would much preferr to laugh with you than at you. I don't care if you laugh at me I been doing it for years. If it makes you laugh and smile I like that its a good thing . But if you want to BS. about unknowns I can sling it as well as the next guy.

[Killnine]

Ever since Conroe came out it been this.

1) spring IDF . Intel rigged the results.

2) Its really not fair to test graphics at low res because its not real world.

3) Its not that big of a lead only 20 to 30%

4) wait till K8l comes out it beat conroe by 40%

5) Torrenza will be the all everthing. Now even AMD is backing intel IBM solution . on the PCI-E

6) Csi doesn't have the bandwidth that direct connect HT3.0 has.

7) AMD still owns the server market . The highend is Intels Itanic the DP is Intels and with Tiger the server market will be intel.

I could go on and on . I could put tons of BS in my sig. But I would much preferr to laugh with you than at you. I don't care if you laugh at me I been doing it for years. If it makes you laugh and smile I like that its a good thing . But if you want to BS. about unknowns I can sling it as well as the next guy.


Nice man, nice.

I am sure AMD will have a kickass solution. But personally I feel like they should dominate with the time they ahve been spending getting their product to market. Its just like the G80 and R600(and the flamewar comes full circle!),
while R600 may outperform G80 by some margin, IT SHOULD for the fact that it is coming out MONTHS after G80.

However, at this point (and seeing that I will be purchasing my upgrades in Jan.), I see Kentsfield as the processor I want, simply because I dont want two friggin processors. Just seems kinda backwards to me. I mean, we might as well start talking about Tigerton and INTELS multiprocessor, multicore solution.

Frankly, I want to see what single die processors can do. (and that also goes for GPUs, in my case =D)

[Turtle 1]

Nice man, nice.

I am sure AMD will have a kickass solution. But personally I feel like they should dominate with the time they ahve been spending getting their product to market. Its just like the G80 and R600(and the flamewar comes full circle!),
while R600 may outperform G80 by some margin, IT SHOULD for the fact that it is coming out MONTHS after G80.

However, at this point (and seeing that I will be purchasing my upgrades in Jan.), I see Kentsfield as the processor I want, simply because I dont want two friggin processors. Just seems kinda backwards to me. I mean, we might as well start talking about Tigerton and INTELS multiprocessor, multicore solution.

Frankly, I want to see what single die processors can do. (and that also goes for GPUs, in my case =D)

Interesting no E-penis cool

[grimREEFER]

I don't know if you can say that about grimreefer...I don't think anyone with a smithfield in their sig can possibly be an Intel fanboy :lol:

Perkam
lol

[Theli]

However, at this point (and seeing that I will be purchasing my upgrades in Jan.), I see Kentsfield as the processor I want, simply because I dont want two friggin processors. Just seems kinda backwards to me. I mean, we might as well start talking about Tigerton and INTELS multiprocessor, multicore solution.

Frankly, I want to see what single die processors can do. (and that also goes for GPUs, in my case =D)
Same here, I have no intentions of buying a multiprocessor system as cooling it would be too much work. Installing one Scythe Ninja was enough of an adventure for me, it cut my fingers up quite nicely.

And in my experience, the best people to ask for advice when it comes to upgrades are the people who are in the market themselves and who actually need CPU performance (assuming I do also), as they cannot afford to be unrealistic, intellectually dishonest or partisan. The worst person to ask would have to be one sitting on a system that has just recently become "old".

[ozzimark]

Ever since Conroe came out it been this.

1) spring IDF . Intel rigged the results.

2) Its really not fair to test graphics at low res because its not real world.

3) Its not that big of a lead only 20 to 30%

4) wait till K8l comes out it beat conroe by 40%

5) Torrenza will be the all everthing. Now even AMD is backing intel IBM solution . on the PCI-E

6) Csi doesn't have the bandwidth that direct connect HT3.0 has.

7) AMD still owns the server market . The highend is Intels Itanic the DP is Intels and with Tiger the server market will be intel.

I could go on and on . I could put tons of BS in my sig. But I would much preferr to laugh with you than at you. I don't care if you laugh at me I been doing it for years. If it makes you laugh and smile I like that its a good thing . But if you want to BS. about unknowns I can sling it as well as the next guy.
it does make me laugh actually.... :D:D

two things i want to address though ;)
2. i won't call it unfair, but it's a good point. you buy a cpu to game on, even if the cpu is 19057208672x faster, if you're gpu limited in the first place...... :D
which sorta brings us back on topic. in my eyes, 4x4 really is a waste for most people. games aren't multithreaded enough yet... though i do suppose that the CPU overhead of running 4 cards in parallel will require some extra cpu oompf, and nvidia's drivers ARE multithreaded iirc.. so while a poor value, it really will pump out power. will it beat yorkfield? for the price, definitely not. in overall performance.. we'll see. definitely not in single threaded apps, that's for sure.

4. who has said that revH be 40% faster than conroe/yorkfield? i've heard 40% over current K8 performance, but not 40% over what intel has to provide. if you can show me where someone has said what you did, i'll gladly shut up and get out of your face :toast:

[Theli]

4x4 really is a waste for most people. games aren't multithreaded enough yet...
I'm sure the 4x4 isn't marketed for "most people" anyway, so I think we can pretty much them out directly. :p:
I'm also pretty sure that most computer owners have rather old computers (think 800MHz) and couldn't run the latest games anyway. And most people in the world have no computers at all.

Question is... will it be good for enthusiasts? Overclockers? People who don't like to wait long for an image or animation to render?
This is why I asked about TDP, will it be feasible to run an 8-core 4x4 system overclocked on air? Will the fact that you have to have two processors running at the same clock frequency hinder overclocking?

[ozzimark]

hehe, yeah. it really is a limited market... :p:

for overclocking. for 8 cores.. dual quad core, remember that revH should allow you to change the multi for each core. one core performs way worse than the other 7? drop the multi a notch :woot:

and if you have a well ventilated case, heat concerns shouldn't be too different than a single socket solution ;)

[Turtle 1]

it does make me laugh actually.... :D:D

two things i want to address though ;)
2. i won't call it unfair, but it's a good point. you buy a cpu to game on, even if the cpu is 19057208672x faster, if you're gpu limited in the first place...... :D
which sorta brings us back on topic. in my eyes, 4x4 really is a waste for most people. games aren't multithreaded enough yet... though i do suppose that the CPU overhead of running 4 cards in parallel will require some extra cpu oompf, and nvidia's drivers ARE multithreaded iirc.. so while a poor value, it really will pump out power. will it beat yorkfield? for the price, definitely not. in overall performance.. we'll see. definitely not in single threaded apps, that's for sure.

4. who has said that revH be 40% faster than conroe/yorkfield? i've heard 40% over current K8 performance, but not 40% over what intel has to provide. if you can show me where someone has said what you did, i'll gladly shut up and get out of your face :toast:

Check post by Serge or Ghost. In those threads you find read everyone else's.

You aren't in my face. If your good with something it ok by me.

[The Ghost]

Check post by Serge or Ghost. In those threads you find read everyone else's..
not by me , i have not said that K8L will be 40% faster then conroe based cpu's

[ozzimark]

Check post by Serge or Ghost. In those threads you find read everyone else's.
i did, they are all referencing up to 40% improvement over current K8... i wouldn't ask you to show something if i haven't already looked :toast:

[freecableguy]

http://www.thetechrepository.com/showthread.php?t=56

[informal]

http://www.thetechrepository.com/showthread.php?t=56

FCG i read that post of yours and I have one question: To someone who is going to do heavy multitasking with many instances of audio/video enc. and rendering at the same time,what do you think how would one lowly 6300(even sky high clocked) cope with those 5+ tasks and still obtain responsivness of the whole system?
Let me answer that one: It CAN'T!

Enthusiast won't buy 6300 at the first place.They would buy 6600(or up).But you end up with the same shared cache wich would trache like hell in 5+ tasks which won't fit in it's L2(all tasks combined).Not to mention the cooling you need since C2D gets really really hot when you pass certain voltage and freq. mark(at least really hot for the 65nm chip)
So despite the fact that you make some valid points in that post,i can't say i understand why you put down AMD so much...All this after many confirmations that the platform(4x4) supports Altair Quad Core FX chips which will be a large step forward in arch . terms,4 Graphic cards,probably socket F accelerators when they show up(H.Richards spoke about this in DigiTimes interviews),non regged memory which means enthusiast high speed DDR2 models are a must for those interested.
Intel has a better core at the moment but lacks better interconnects which will be AMD's main weapon until New Core arrives (and probably crush the competition-my hopes are at least).

[sierra_bound]

@informal

There are many members here with 6300 & 6400 Allendales. Are you saying they are not enthusiasts? :confused:

I think what FCG is saying is that 4X4 won't be cheap. In order to get dual-channel performance you will need four memory modules. Of course, Kentsfield won't be cheap either. But it's compatible with existing Conroe boards, so you if you already own C2D, you won't be forced to buy a new board (and maybe new memory).

There are fanboys on both sides of the aisle. What I've observed about the Intel vs. AMD debate is reminiscent of the lyrics from an old Simon & Garfunkel song. "All lies and jest, still the man hears what he wants to hear and disregards the rest."

[informal]

@informal

There are many members here with 6300 & 6400 Allendales. Are you saying they are not enthusiasts? :confused:

I think what FCG is saying is that 4X4 won't be cheap. In order to get dual-channel performance you will need four memory modules. Of course, Kentsfield won't be cheap either. But it's compatible with existing Conroe boards, so you if you already own C2D, you won't be forced to buy a new board (and maybe new memory).

There are fanboys on both sides of the aisle. What I've observed about the Intel vs. AMD debate is reminiscent of the lyrics from an old Simon & Garfunkel song. "All lies and jest, still the man hears what he wants to hear and disregards the rest."

No not at all.They are enthusiast alright.I just meant those people that fit in narrow group and who multitask a LOT.Thats all.The ones who need the power of 4 cores...Truth be told,there aren't many of them.

[Bloody_Sorcerer]

every single DC addict needs 4 cores... well, we really need as many cores as we can get our grubby hands on ;)

more on the topic:

<insert random stupid fanboyist bull:banana::banana::banana::banana: here about future products with absolutely no evidence or support or anything based in the realm of facts>

TRUTHINESS FOR THE WIN.

[crackmann]

what ever,
as soon as AMD has a direct HT link from the CPU(s) (or each core) to the Videocards they will win.

p.s. intel is dumb

[[XC] flat-four]

TRUTHINESS FOR THE WIN.

If thats not a Steven Colbert-ism, I don't know what is.

Anyway, 4x4 is a DC addicts wet dream. The ability to have 2xquad core (in 12 months) with enthusiast memory is an amazing thing. Same goes for most of the graphics design industry that doesn't really need registered memory. Those tight timings and high speed would rip through some hard core rendering. I think intel, while woodcrest is very tempting, is missing out by limiting dual socket workstations to require fb-dimms. Just think about dual 5160's(or some 5300 series clovertowns) running ddr2@1000+ with a healthy overclock. This is exactly the market that AMD is aiming for. Dual k8l running ddr2@1000+ is a beautiful thing. I'm saving up my nickels and dimes for a nice 4x4 rig.

[pH(x)]

So if I get the message right, Intel is pushing for ECC RAM for it's 2-socket system, yet AMD is not?

Quite a performance advantage for AMD, plus the price difference over ECC is a huge incentive to go for the late-comers.

[Canibuz]

I heard of HTX months before CSI.

[sierra_bound]

So if I get the message right, Intel is pushing for ECC RAM for it's 2-socket system, yet AMD is not?

Quite a performance advantage for AMD, plus the price difference over ECC is a huge incentive to go for the late-comers.
Again, remember with 4X4 you will need two of everything. More thank likely, two sets of RAM if you want to run dual-channel, and two HSF's or two waterblocks. So there really is no savings.

[VulgarHandle]

Again, remember with 4X4 you will need two of everything. More thank likely, two sets of RAM if you want to run dual-channel, and two HSF's or two waterblocks. So there really is no savings.
cause most servers don't have 4 gigs+? two HSF's? and screw water, single phase with heatipipes to blocks, perfect, since i'm sure we can count on cold-bug for of course 90nm, and probably the same if not worse on 65nm

try and understand the platform a little more, and hooray for 2P power for less moolah and more bang

[sierra_bound]

cause most servers don't have 4 gigs+? two HSF's? and screw water, single phase with heatipipes to blocks, perfect, since i'm sure we can count on cold-bug for of course 90nm, and probably the same if not worse on 65nm

try and understand the platform a little more, and hooray for 2P power for less moolah and more bang
What does this have to do with servers? I seem to recall AMD executives saying that 4X4 was designed to counter Core 2 Duo. Also, I'm not interested in servers.:) And I'm not an AMD fanboy. I use both AMD and Intel. In fact, I'm typing this message on an Opteron system. I don't have blind allegiance to one or the other like some people.

The point I was making earlier in this thread is that some people will use twisted logic to support either AMD or Intel. I've never viewed it as "either, or". Also, I don't look the other way when either AMD or Intel does something stupid.

And "more bang" you say? I'm sure you've tested both 4X4 and Kentsfield and can enlighten us all on which is better.

[duploxxx]

What does this have to do with servers? I seem to recall AMD executives saying that 4X4 was designed to counter Core 2 Duo. Also, I'm not interested in servers.:) And I'm not an AMD fanboy. I use both AMD and Intel. In fact, I'm typing this message on an Opteron system. I don't have blind allegiance to one or the other like some people.

The point I was making earlier in this thread is that some people will use twisted logic to support either AMD or Intel. I've never viewed it as "either, or". Also, I don't look the other way when either AMD or Intel does something stupid.

And "more bang" you say? I'm sure you've tested both 4X4 and Kentsfield and can enlighten us all on which is better.

well more bang for the buck has to be seen, we'll know i a few weeks. It will produce more heat and is a new mobo.
But for sure we know that socket F 2p is 5-10% slower than equal 2p woodcrest at same clock on 32bit. Knowing that Kentsfield is only 1066fsb and the best asus board gives you 8x8x4x8 on current chipsets with no space in between for big gpu cooling. Against a full speed (no bus limit = opteron) and 16x8x16x8 system with less space restriction with normal memory so again better performance. I'll put my bet on the QF for sure... future proof and 8core ready (intel will go to 1333 in next chipset release)

[sierra_bound]

well more bang for the buck has to be seen, we'll know i a few weeks. It will produce more heat and is a new mobo.
But for sure we know that socket F 2p is 5-10% slower than equal 2p woodcrest at same clock on 32bit. Knowing that Kentsfield is only 1066fsb and the best asus board gives you 8x8x4x8 on current chipsets with no space in between for big gpu cooling. Against a full speed (no bus limit = opteron) and 16x8x16x8 system with less space restriction with normal memory so again better performance. I'll put my bet on the QF for sure... future proof and 8core ready (intel will go to 1333 in next chipset release)
Okay that makes more sense than the other person's argument. But the bottom line is that if 4X4 does not equal or exceed the performance of Kentsfield, no one is going to buy it except AMD diehards. A lot of people have already invested $$$ in Conroe systems, which as you probably know, are Kentsfield compatible. Those people aren't going to spend more $$$ to switch unless there is a compelling reason.

[VulgarHandle]

What does this have to do with servers? I seem to recall AMD executives saying that 4X4 was designed to counter Core 2 Duo. Also, I'm not interested in servers.:)
which is why i said try and understand the platform better, it's not gonna beat c2d at superpi, or many other single-threaded apps, it's meant to do much more, it's like Bush using WMD's as the reason to go into Iraq, when there were many more better reasons to go

And I'm not an AMD fanboy. I use both AMD and Intel. In fact, I'm typing this message on an Opteron system. I don't have blind allegiance to one or the other like some people. The point I was making earlier in this thread is that some people will use twisted logic to support either AMD or Intel. I've never viewed it as "either, or". Also, I don't look the other way when either AMD or Intel does something stupid.
didn't say u did, never even thought it, and glad to hear it, but calling this stupid is a little slanted

And "more bang" you say? I'm sure you've tested both 4X4 and Kentsfield and can enlighten us all on which is better.
the more bang for less cash had to do with comparing the price against a 2p 2200 series opteron system

But the bottom line is that if 4X4 does not equal or exceed the performance of Kentsfield, no one is going to buy it except AMD diehards.
yeah, of course, as long as the tests used to determine it aren't limited to a single run of superpi or 3dmark01

[sierra_bound]

the more bang for less cash had to do with comparing the price against a 2p 2200 series opteron system
The Opteron 2210 is only around $260. Not bad. The 2220 is very expensive.

yeah, of course, as long as the tests used to determine it aren't limited to a single run of superpi or 3dmark01
It doesn't matter which tests are used. It's the perception that counts. If Kentsfield is perceived to be faster, AMD will lose. Lots of Kentsfield tests with multi-threaded programs like Cinebench have already been posted at this forum. 4X4 is still somewhat of an unknown. I'm sure we will learn more in the coming weeks.

[VulgarHandle]

The Opteron 2210 is only around $260. Not bad. The 2220 is very expensive.
2218 is 2.6ghz, but $661 FOR ONE, which is probably the price of the FX70 SKU(2.6ghz), which is for both(estimated based on latest info saying the FX74 would match the price of the QX6700, which was said to list for $999 at launch)

[Turtle 1]

http://www.thetechrepository.com/showthread.php?t=56

Again good job. I wish I had your youth and your ability to communicate;)

[nn_step]

Yet I don't think everyone here gets how HUGE this is going to be. It is going to Kill Intel's 2P market

[uOpt]

http://www.thetechrepository.com/showthread.php?t=56

Good article.

However, I think you overestimate the effect of NUMA. In all the years since Opterons came out (even the first ones had NUMA on most mainboards) I have not seen a benchmark showing a real-world advantage in significant ranges (> 3%), except for one dedicated SQL server benchmark.

I think for 4 cores on a mixed-purpose system none of the NUMA, Hypertransport, quad-channel and whatever else things matter.

[uOpt]

So if I get the message right, Intel is pushing for ECC RAM for it's 2-socket system, yet AMD is not?

Quite a performance advantage for AMD, plus the price difference over ECC is a huge incentive to go for the late-comers.

You mix up registered and ECC.

All these platforms can use but don't require ECC. But Intel 771 chipsets and Socket F require registered and 4x4 seems to require or be able to use unregistered.

[arisythila]

i did, they are all referencing up to 40% improvement over current K8... i wouldn't ask you to show something if i haven't already looked :toast:

LOL... Tables are turning on Turtle............. :stick: :fact:

[arisythila]

yeah, of course, as long as the tests used to determine it aren't limited to a single run of superpi or 3dmark01

I seriously cannot believe that people still use 3dmark01... Seems like a lost cause to me. IF anything I want to see some 3dmark06 marks, maybe some 05. But 01.... :banana::banana::banana::banana: man, that crap is almost 7 years old.

(Yes, were gonna break 100k)
[f*ck 100k man, I want to see 15k in 3dmark06]

Know what i mean?

~Mike

[red]

Kill Intel's DP? Clovertown will be here at least 6 months before Barcelona, at least 9 months before quad core FX.

From his list of 7, only disputing one is pretty good:)

[Turtle 1]

Ya I guess I should retract that statement, and use 20% faster than Conroe which is still laughable.

I also need to add 8) Conroe won't show to the masses till Sept maybe Oct.

Look guys I already apologized for being an Intel fanboy . I wish I wasn't . But I is what I is. and thats all there is to it. If Intel went out of business and my only choice was AMD . I would walk away from my screen and never return.

[Poodle]

Ya I guess I should retract that statement, and use 20% faster than Conroe which is still laughable.

I also need to add 8) Conroe won't show to the masses till Sept maybe Oct.

Look guys I already apologized for being an Intel fanboy . I wish I wasn't . But I is what I is. and thats all there is to it. If Intel went out of business and my only choice was AMD . I would walk away from my screen and never return.


:rofl: It's official: Turtle is the biggest fanboy in space! :fact:

With respect again (it's hard I know, but I try): Why do you even bother to post in this forum? No one here takes a fanboy seriously. You're wasting your time. :stick:

[red]

But I is what I is.
:rolleyes:
At least you can capitalize;)

[Turtle 1]

Thanks poodle that made me smile and life . Feels good don't it?

[Poodle]

Thanks poodle that made me smile and life . Feels good don't it?

Yeah, it's good fun :) Can't get though how you can be bothered posting all this stuff that you actually are, when you know that we are not buying? :confused: Most people that hang out here know a fanboy statement when they see one, you know. This is not Hardforum, Thg or some other lame forum...

[Lightman]

Again, remember with 4X4 you will need two of everything. More thank likely, two sets of RAM if you want to run dual-channel, and two HSF's or two waterblocks. So there really is no savings.

WHAT A MISTAKE! You're copying FCG's false claims! Are you ever looked at memory performance of Opteron based 2P/4P servers? Small TIP: look at SiSoftSandra database <wink>.

To clarify: 4x4 or any 2P system on Opteron will give you QUAD Channel memory performance if you will use DC mode on both sockets! That's when NUMA is used!

Cheers!

[Turtle 1]

Yeah, it's good fun :) Can't get though how you could be bothered posting all this stuff that you actually are, when you know that we are not buying? :confused:

Well its hard. When you discuss hardware thats not available to test.

We need to research about it than we Post about it. Some of the stuff is way out in left field. People shouldn't get mad because some of the stuff you research is incorrect. Best thing is to laughed it off . Defend your position right or wrong until its proven wrong. Than give up the Ship.

Take Tigerton many reports are saying point to point from the Cpu to the NB along with 4 independent FSB .Until yesterday I have never heard anyone referr to a FSB as a point to point connection.
Doing so is Wrong and that can be proven. So if these sites are reporting a point to point connection from the Cpu to the N/B and they are referring to the 4 dependent FSB that just bad reporting and its false.

[Theli]

But I is what I is.
:rolleyes:
At least you can capitalize;)
The most irritating rape of the english language by far is when people type "u" instead of "you", or "u'll" instead of "you'll". They can't be bothered to type two extra letters?

[Turtle 1]

Here some old stuff about futere stuff.The fact that intels shows actuall working models says alot about the company wheather or it makes it to mainstream thats another story.

http://www.anandtech.com/showdoc.aspx?i=2511&p=1

[Turtle 1]

Yeah, it's good fun :) Can't get though how you can be bothered posting all this stuff that you actually are, when you know that we are not buying? :confused: Most people that hang out here know a fanboy statement when they see one, you know. This is not Hardforum, Thg or some other lame forum...

Your right this isn't [H] or AT or other lame forums.

I don't post much in the other sections unless I see a balant fanboy post from a AMDroid. That has zero creditability. Than I post. I read the Most important section here the Intel section. But mostly I read their . Most those guys are out of my league. Plus I am not an exstreme O/C . Instead I concentrate on a stable O/C . Than I work to slap around other peoples benchies at the same clock.

[Gag3]

Ya I guess I should retract that statement, and use 20% faster than Conroe which is still laughable.

I also need to add 8) Conroe won't show to the masses till Sept maybe Oct.

Look guys I already apologized for being an Intel fanboy . I wish I wasn't . But I is what I is. and thats all there is to it. If Intel went out of business and my only choice was AMD . I would walk away from my screen and never return.

I bet you hated the fact that Intel had to copy AMD inorder to beat them.

You may aswell walk away from your screen now.. K8L is going to be the most powerful desktop processor, like it or not.

[Turtle 1]

well more bang for the buck has to be seen, we'll know i a few weeks. It will produce more heat and is a new mobo.
But for sure we know that socket F 2p is 5-10% slower than equal 2p woodcrest at same clock on 32bit. Knowing that Kentsfield is only 1066fsb and the best asus board gives you 8x8x4x8 on current chipsets with no space in between for big gpu cooling. Against a full speed (no bus limit = opteron) and 16x8x16x8 system with less space restriction with normal memory so again better performance. I'll put my bet on the QF for sure... future proof and 8core ready (intel will go to 1333 in next chipset release)


On the Asus 8x8x4x8 . since this is XS 4 small water blocks allows for a lot of room. Your also in the news section and should be aware that bearlack chipset is coming. Same as K8L is coming. The Bearlake chipset is also PCI-e2. So why is it Amdroids say its not fair to compare Conroe to outdated K8 stuff. Than they turn around and compare future stuff from AMD on present stuff from Intel.

Wouldn't it be more proper to compar future stuff from AMD to future stuff from intel do out in the same time frame. At least I admit what I am.

[Turtle 1]

I bet you hated the fact that Intel had to copy AMD inorder to beat them.

You may aswell walk away from your screen now.. K8L is going to be the most powerful desktop processor, like it or not.

So what has intel copied from AMd. AMD64. Ya 3 years old and not much use yet. K8L seems to have copied a few things from conroe. wouldn't you say thats true.

[Ailleur]

I bet you hated the fact that Intel had to copy AMD inorder to beat them.

You may aswell walk away from your screen now.. K8L is going to be the most powerful desktop processor, like it or not.

Boy, im sure you have some solid diagrams, simulations, or any concrete numbers that back that claim.

The X factor, aka "i pulled it out of my ass" "or "i extrapolated numbers from AMD presentation powerpoint slides with pretty colors i saw on the web" is not considered solid.


As for "copying", how would YOU know what was copied or not? You probably dont have the slightest clue as to how they both work internaly unless you actually work for both companies at the same time.
Thats like saying ikea copied the idea of "making furniture" from someone else.

[Turtle 1]

But I is what I is.
:rolleyes:
At least you can capitalize;)


Red you know were I got that from . 45 years back ,

I is what I is um Popeye the sailor man tot tot.

[zakelwe]

well more bang for the buck has to be seen, we'll know i a few weeks. It will produce more heat and is a new mobo.
But for sure we know that socket F 2p is 5-10% slower than equal 2p woodcrest at same clock on 32bit. Knowing that Kentsfield is only 1066fsb and the best asus board gives you 8x8x4x8 on current chipsets with no space in between for big gpu cooling. Against a full speed (no bus limit = opteron) and 16x8x16x8 system with less space restriction with normal memory so again better performance. I'll put my bet on the QF for sure... future proof and 8core ready (intel will go to 1333 in next chipset release)

"Knowing that Kentsfield is only 1066fsb and the best asus board gives you 8x8x4x8 on current chipsets with no space in between for big gpu cooling."

Err ..

http://service.futuremark.com/compare?2k3=4915655


Lets see how fast 4x4 is by how many top 10 Futuremark Orb places it has come January 1st 2007. That's always a guide to how fast a cpu is and how good it will be for gaming.

Regards

Andy

[arisythila]

GAY. I want to see some benchies w/ atlease 1600x1200 with AA 8X, and AF 16x... I dont want to see complete TRASH!

3dmark05, or 06 only please.

Anything else renders complete trash. (Other 3dmark's)

~Mike




"Knowing that Kentsfield is only 1066fsb and the best asus board gives you 8x8x4x8 on current chipsets with no space in between for big gpu cooling."

Err ..

http://service.futuremark.com/compare?2k3=4915655


Lets see how fast 4x4 is by how many top 10 Futuremark Orb places it has come January 1st 2007. That's always a guide to how fast a cpu is and how good it will be for gaming.

Regards

Andy

[Anonymous]

I bet you hated the fact that Intel had to copy AMD inorder to beat them.

You may aswell walk away from your screen now.. K8L is going to be the most powerful desktop processor, like it or not.


Well since most of the Xtreme News lurkers are from the future maybe you can bring back the "most powerful desktop processor" to the year 2006 and show us all :rolleyes:

[Poodle]

So what has intel copied from AMd. AMD64. Ya 3 years old and not much use yet. K8L seems to have copied a few things from conroe. wouldn't you say thats true.

Good going there! :)

But you have to enlighten me please. :)


I will buy that the Conroe release made Hector running around the office setting fire to every one's asses shouting: "we have to meet this performance with something really quick" but copied, I think not... They made the fpu bigger and such to stand up to c2d but there the similarities end. Intel and Amd don't have the same Engineering tradition. Like comparing US weapons with Russian. American guns are more refined but Russian weapons never click and have some out of the box solutions. Like shooting missiles out of the barrel of a tank as russian missiles don't have wings. American missiles have wings and thus can't be fired like that though they can do alot that Russian weapons can't. None is better overall, just different.

K8L don't have a FSB or shared L2. It has shared L3 and is a native core (e.g. compared to Kenty which has/is neither). The Engineers have made different choses.

I'm not saying any is the better. I just don't think they are much alike. C2D will probably cost less and thus be better, or something. Who knows.

[Turtle 1]

Good going there! :)

But you have to enlighten me please. :)


I will buy that the Conroe release made Hector running around the office setting fire to every one's asses shouting: "we have to meet this performance with something really quick" but copied, I think not... They made the fpu bigger and such to stand up to c2d but there the similarities end. Intel and Amd don't have the same Engineering tradition. Like comparing US weapons with Russian. American guns are more refined but Russian weapons never click and have some out of the box solutions. Like shooting missiles out of the barrel of a tank as russian missiles don't have wings. American missiles have wings and thus can't be fired like that though they can do alot that Russian weapons can't. None is better overall, just different.

K8L don't have a FSB or shared L2. It has shared L3 and is a native core (e.g. compared to Kenty which has/is neither). The Engineers have made different choses.

I'm not saying any is the better. I just don't think they are much alike. C2D will probably cost less and thus be better, or something. Who knows.

reported to be 4issue shared cache be it L3 which intel already has.

Snoopy filter Tigerton has that. If AMD adapts sse4 . But AMD won't get that on K8L Intel held it back . Than Intel is only releasing 30 of those on penryn( Already on Merom just not activated .) . Intel is holding more than 20 of those for Nehalem so AMD won't get those on K10 until later.

So what did Intel copy of AMD's? And your right AMD&Intel do not implament the same tech the same . Keep that in mind when Intel comes with csi and ondie memory controller

[Turtle 1]

intel already has this....

intel are holding back the SSE4 license, something you can easily call illegal as they have a cross license agreement with amd - whatever intel includes in a cpu and can as well, and the oposite

even if intel is silent about parts of the instructions, the moment they finalise specs you can call it also amd tech... and there is no such thing as too late - first of all this things take long to adopt by software devs

but thinking AMD needs a new core so it can put sse4 inside is lame. What AMD needs is a simple revision, like the 13nm athlons feature only sse2 but the 9nm also feature

intel has copied a lot in the past years, K8 was a great inspiration for them. amd also copies from intel, thats what their agreement is all about



When? Do you know? Intel show 80 core chips, all kind of stuff, but still no CSI.
Intel does not have to give AMD anything until they implement it.

Did AMD give intel AMD 64 before it was launched and used . noway . Ya I new Amd could implement on revision but K8L isn't going to be out long before K10 arrives. So they wont have alot of time to do a revision On K8L

Your right though Intel does have to share x86tech , but not on till after its implamented on intels cpu's and activated.

But nehalem c is the last time intel will have to share with amd. Geshner 32 core cpu is not going to be X86. It will run x86 software using a compiler based on Elbrus tech. Intel bought Elbrus in 04.

[vitaminc]

I bet you hated the fact that Intel had to copy AMD inorder to beat them.

You may aswell walk away from your screen now.. K8L is going to be the most powerful desktop processor, like it or not.

and AMD had to copy Intel to beat them...

so what? nothing in the intel/amd x86/x86 64 architecture is new anyway, just the matter of when and who put it on x86 architecture first.

[zakelwe]

GAY. I want to see some benchies w/ atlease 1600x1200 with AA 8X, and AF 16x... I dont want to see complete TRASH!

3dmark05, or 06 only please.

Anything else renders complete trash. (Other 3dmark's)

~Mike

You have completely missed the point I was making. I'm not talking about how pretty the ORB is but the correlation between the best cpu for gaming and ORB results.

Regards

Andy

[WeStSiDePLaYa]

you are all heroes

[generics_user]

reported to be 4issue shared cache be it L3 which intel already has.
something that ibm (ok actually freescale in its licensed cpu) already had earlier inside the PPC 7450 intruduced in january 2001;)

[NapalmV5]

But I is what I is.
:rolleyes:
At least you can capitalize;)

why?!? capatilized... wat about You why its not demanded? hmm...

who ever came up with this grammer poop must have been like :I am alpha! I am omega! yes!, I! I! I!:


The most irritating rape of the english language by far is when people type "u" instead of "you", or "u'll" instead of "you'll". They can't be bothered to type two extra letters?

rape english? wtf? i rape any langauage i can.. :D

im deeply sorry for my rant :D i think ive had too much.. or lack of it? uhh yeh...

good night all! :)

[Theli]

rape english? wtf? i rape any langauage i can.. :D

im deeply sorry for my rant :D i think ive had too much.. or lack of it? uhh yeh...

good night all! :)
Yes, go and sleep it off...

[Turtle 1]

Here is how I see what AMD would do, should they Choose MCM
http://img.photobucket.com/albums/v643/nn_step2/explaining.png

Just wanted to save the diagram. As one has already disappeared.

Someone had a drawing based on this link.Saying that the 2 nv nb would connect to seperate cpu's we now know thats false.


http://www.amd.com/us-en/assets/content_type/DownloadableAssets/2P_S_WS_Comparison_PID_41460.pdf

[Fred_Pohl]

if this is true, octa core for less than 1 kentsfield!

It's not true. Even a 4C (2xFX74) K8 4x4 system will cost more than Kentsfield and K8L is still ~6 months away. By the time K8L arrives there will be $500 Kentsfields and I don't imagine AMD will be selling a pair of K8Ls for under $500.

[duploxxx]

Intel does not have to give AMD anything until they implement it.

Did AMD give intel AMD 64 before it was launched and used . noway . Ya I new Amd could implement on revision but K8L isn't going to be out long before K10 arrives. So they wont have alot of time to do a revision On K8L

Your right though Intel does have to share x86tech , but not on till after its implamented on intels cpu's and activated.

But nehalem c is the last time intel will have to share with amd. Geshner 32 core cpu is not going to be X86. It will run x86 software using a compiler based on Elbrus tech. Intel bought Elbrus in 04.

If it was you INtel would bring it all and can do all.....

CSI is comming X86 compiler..... sure.

How many revisions dit it take to bring C2D to current level.... Banias-dothan-yonah-merom..... and you think they can bring CSI to a level that AMD already has for several years (it took them also somw revisions).... sure... they know how to use fsb/cache and workarounds that's it. What's comming next from Intel. the current C2D on a smaller die and some cache....

[cky2k6]

It's not true. Even a 4C (2xFX74) K8 4x4 system will cost more than Kentsfield and K8L is still ~6 months away. By the time K8L arrives there will be $500 Kentsfields and I don't imagine AMD will be selling a pair of K8Ls for under $500.
good job of twisting my words, but i have no idea what youre trying to prove. what i simply stated was that if 4x4 was two quad cores, and amd promised that 4x4 will be under 1000 except for the highest end, therefore you have octa core for less than the competition's quad. doubt its gonna happen though...

[red]

http://www.tgdaily.com/2006/11/10/amd_4by4_details/
*4x4 won't touch the anticipated $800 - $900 price range.
*The "entry-level" 4x4 solution will be sold as a dual-processor package and will sell for about $1000.
*will be upgradeable to quad-core processors, currently code-named "Agena," which are expected to launch sometime in mid-2007.

- From Desktop Marketing Director Leslie Sobon

[nn_step]

it is called investing for Oct core silly

[red]

If you're going to throw away tons of $$$ for an inferior quad solution, you might as well throw away more for the real thing(AM2+) in the future. So I see no point in the upgradeability of 4x4. If you really need 8 cores, there is Clovertown for a good at least 3 quarter lead.

[nn_step]

If you're going to throw away tons of $$$ for an inferior quad solution, you might as well throw away more for the real thing(AM2+) in the future. So I see no point in the upgradeability of 4x4. If you really need 8 cores, there is Clovertown for a good at least 3 quarter lead.
At the same time a good Clovertown board with support for 16GB costs how many arms and legs :stick:

[Turtle 1]

At the same time a good Clovertown board with support for 16GB costs how many arms and legs :stick:

When talking about this kind of $$$ . Whats the differance. Both are going to be expensive. So its just a matter of performance . The top performer should be the one that is choosen on its merits alone and not a few extra bucks. IF you can afford 4x4 you can afford cloverton.

[nn_step]

When talking about this kind of $$$ . Whats the differance. Both are going to be expensive. So its just a matter of performance . The top performer should be the one that is choosen on its merits alone and not a few extra bucks. IF you can afford 4x4 you can afford cloverton.
Yep, now if you are doing Bandwidth dependent applications, (applications with alot of IO) 4x4 has a HUGE edge over Coverton. But if you do alot of Processing with very little IO, Cloverton is a better choice. It depends on what you are going to be doing the most. They both have their merits.

[zakelwe]

http://www.tgdaily.com/2006/11/10/amd_4by4_details/
*4x4 won't touch the anticipated $800 - $900 price range.
*The "entry-level" 4x4 solution will be sold as a dual-processor package and will sell for about $1000.
*will be upgradeable to quad-core processors, currently code-named "Agena," which are expected to launch sometime in mid-2007.

- From Desktop Marketing Director Leslie Sobon

That slide they show says Crossfire as well as SLI for this .. however it is an nvidia chipset ? :confused: Does that mean that there will be an Ati/AMD chipset version of 4x4 as well then at some point ?

Regards

Andy

[el rolio]

yea they annouced/mentioned it earlier this week

[The Ghost]

http://www.tgdaily.com/2006/11/10/amd_4by4_details/
*4x4 won't touch the anticipated $800 - $900 price range.
*The "entry-level" 4x4 solution will be sold as a dual-processor package and will sell for about $1000.
*will be upgradeable to quad-core processors, currently code-named "Agena," which are expected to launch sometime in mid-2007.

- From Desktop Marketing Director Leslie Sobon
you know what ? it isn't even worth arguing about if amd top performer 4x4 is going to come out at a $1,000 or less , like amd sais , who's word are we going to take / it doesn't make any difference , by 5 days or less we will know what it costs

[Fred_Pohl]

good job of twisting my words, but i have no idea what youre trying to prove. what i simply stated was that if 4x4 was two quad cores, and amd promised that 4x4 will be under 1000 except for the highest end, therefore you have octa core for less than the competition's quad. doubt its gonna happen though...

Apparently you meant something other than what you said. Please forgive my inability to inuit what you meant, but failed to convey in words.

I'm not trying to prove anything. Just stating the fact that 4x4 will launch with 90nm dual cores and there won't be any quad core 4x4 CPUs until Q2-07 at the earliest.

Further, 4x4 may turn out to be a great home workstation platform but it won't be a bargain by any stretch. Mobos will most likely start at >$300 and the cheapest 4x4 CPU (2.6GHz FX70) is slated to cost $1000 for a pair with the top end FX74 (3.0GHz) slated to cost $1500 a pair. AMD can change their FX7x pricing at any time but it would be foolish to expect 4x4 to be a more economical alternative to Kentsfield. Even more foolish would be expecting a 8C 4x4 system to cost less than Kentsfield. It ain't gonna happen for so many obvious reasons that I won't insult your intelligence by listing them.

[SoF]

well reading the first site should be enough - 75% of the ppl don't have the money for such a multi-chip system (me included), 20% do not need 4 cores as 2 are way enough (unless you're in DC)...really no idea who needs such system, not even here on XS as overclocking 8 cores will be hell :rolleyes:
Really can't follow the way AMD is going ^^

[The Ghost]

20% do not need 4 cores
i think that it would be more like 90% do not need four cores

what is there 5% of the programs out there that can even use 4 cores

[StyM]

http://www2.gol.com/users/fcchapel/images/a64/L1N64-SLI_Deluxe.jpg
http://www2.gol.com/users/fcchapel/images/a64/L1N64_LCD.jpg
http://www2.gol.com/users/fcchapel/images/a64/L1N64_case.jpg
http://www2.gol.com/users/fcchapel/images/a64/L1N64_hdd.jpg



- FX (Altair): 2.7GHz~2.9GHz 65nm、Q2/Q3 2007

- FX-76: 3.2GHz, 1MBx2, 90nm, TDP=125W, S1207, Q2 2007
- FX-74: 3.0GHz, 1MBx2, 90nm, TDP=125W, S1207
- FX-72: 2.8GHz, 1MBx2, 90nm, TDP=125W, S1207
- FX-70: 2.6GHz, 1MBx2, 90nm, TDP=125W, S1207

:D

[Tenro]

Nice Pics!


FX (Altair): 2.7GHz~2.9GHz 65nm、Q2/Q3 2007

But, damn, AMD's processors@65nm will be a full year behind Intel's. :(

Well, let's hope that that the new 90nm's are fast... :shrug:

[Vengure]

Yea we can only hope. I decided to wait till january to upgrade so I trying to figure out which processor will be best to pick up then :)

[Turtle 1]

hm, that is a good point, i'm not entirely sure how video data is sent around with sli and two chipsets... i would PRESUME that amd/nvidia are smart enough to take advantage of both links, if possible.. but you're right, it might not simply due to the nature of how sli works. thanks for pointing that out, i'll look into it a bit closer. of course, FCG could always comment on it, since i know he knows a lot about sli/cf :toast:

Turtle 1, again with this memory controller talk. mind explaining in a bit further detail what you mean: "Funny thing is Intel has until K8L to change the NB memory controller."
what would they change about it?

Change the memory controller to work with DDR3 rather than FBDimms. I would think Intel can pull that off . Than woodcrest/yorkfield could have independent point to pointFSB. With york field 12 mb of shared cache between 4 cores would be formidable. I am just as excited about yorkfield as amders are about K8L . Either one in 4x4 would be overkill. But some will do it just for the experiance. Can you imagin the PCmark 05 scores running 4 or more raptors in raid0 . Either system will kick ass. I would preferr 4x4 Intel . But thats not saying 4x4 K8L wel not kick butt.