AMD Phenom CPUs to See Further Delays

[malware]

AMD has recently notified its partners that the launch of higher-end quad-core Phenom processors, including the 9700 and 9900, will be postponed to the second quarter of 2008 from the original schedule of early 2008, according to DigiTimes. However, whether AMD's triple-core Toliman series CPUs will also see delay will be the key decision for the company, since Toliman offers a high price/performance ratio compared with Intel's quad-core CPUs. A delay for Toliman will hurt AMD the most. The reason for the delay of 9700 and 9900 is because AMD has not yet been able to solve the translation lookaside buffer (TLB) erratum found in the chips.

View at TechPowerUp Main Site

 

[UnXpectedError]

yet another blow to amd.. they really seem to be way off the ball lately

 

[btarunr]

I guess this is to synchronise with Intel's delay of the cheap Yorkfield parts. What is wrong with AMD. They're a 'once bright schoolkid now turned junkie'. With Intel delaying it's YF parts this would've a nice time for them to launch the 9700 and 9900. I feel sad for AMD.

 

[sam0t]

I have grown numb to these bad news from AMD. Next time I read the news from AMD is either their bancrupcy or their next killer product.

 

[GLD]

Well pooh! I have been been staying away from the AM2's holding out for the AM2+'s. With this news and the current Phenoms and their L3 bug, the prices of the 690G boards and DDR2 800 are looking pretty affordable. I might just have to go the AM2 route (for now) with a Black Edition X2. Regardless, I won't be going to the blue camp.

 

[NU(GFX)T]

I guess this is to synchronise with Intel's delay of the cheap Yorkfield parts. What is wrong with AMD. They're a 'once bright schoolkid now turned junkie'. With Intel delaying it's YF parts this would've a nice time for them to launch the 9700 and 9900. I feel sad for AMD.

I think Intel's dealying of Yorkfield parts is not due to technical difficulties like AMD. We have already seen reviews of the highest clocked FSB1600 Yorkfield with unlocked multi in healthy status. There shouldn't be any problems with lower frequency locked versions. I see Intel's move as a "holding back the release" more than anything else. Though it might give intel time to improve their TDP on Yorkfields even more and manage to go over 3.2 without stretching the TDP over 130-136.

IMHO even if AMD did release the 9700 and 9900 Intel would be waiting on the fence to throw budget locked-multi Yorkfirlds at them right away and spoil their party.

I hope AMD put their act together eventually and become competitive again.We are already seeing the consequences, Intel's strategic move to delay YFs. I wanted to replace my q6600 with a new YF champion. Now they are holding the new tech out on us.

Same goes for nVIDIA and ATi, nVIDIA trying to shove another dual GPU card down our throats because ATi is not ready with their new architecture and is going to release a dual GPU card based on their current base architecture making nVIDIA not bothering with speeding up the release of G80's successor.

 

[btarunr]

^ Well said lad.

 

[NU(GFX)T]

Interesting. Motherboards are to be blamed for YF delays!

LINK

Well my mobo uses six layer PCB most others that I know also use six layer PCB. I am not sure about abit IP35pro though I couldn't find any infos on its PCB layers anywhere on the net.

 

[pt]

maybe intel and amd talked so they can have a rest for a while and work on their chips so they can perform better
(it's xmas after all )

 

[Deleted member 3]

(it's xmas after all )

Wasn't the core architecture based on the Israeli designs? Happy Hanukkah around this time next year mate.

 

[Cold Storm]

I am with you all on the fact that its the tech. part of the both processor groups being the reason why the delay. But, i feel that both AMD and Intel are working together on this. One is delayed and the other isn't ready ether? I think that they are waiting to put the processors out when Intel put theirs out. Make people wait, and then have the "war" that is needed to ether put them in a better spot or not.
You have to have some type of competition going on. I know, I know, its better to put out something and have time to get the market flooded with your product, but Competition is what fuels the fire to make stuff better..

 

[maxtoons.com]

That is it

That is it, I am going Intel.
I have been running my Optiron 185 for over a year now waitting an answer from AMD for a quad core.
I want to support the little guy, but hey I need a new system for 2008

 

[Deleted member 24505]

that tlb error is sure biting them on the ass.Shame they nvr picked it up earlier.Can they fix it without redoing the chip from scratch? if the error is because of a design fault,its gonna be a doozy to fix.

 

[mandelore]

Though it might give intel time to improve their TDP on Yorkfields even more and manage to go over 3.2 without stretching the TDP over 130-136.

actually, 130WTDP is an over estimation of the thermal output, its just put in that bracket for oem retailers to offer correct/appropriate heatsinks. I believe some tests were done on some 130W yorkies and its was merely 100-110W on stock clocks, which is nice to know

 

[pt]

that tlb error is sure biting them on the ass.Shame they nvr picked it up earlier.Can they fix it without redoing the chip from scratch? if the error is because of a design fault,its gonna be a doozy to fix.

in next revision it will be fixed, and it's very hard to catch the problem, almost no need to use the fix

 

[NU(GFX)T]

actually, 130WTDP is an over estimation of the thermal output, its just put in that bracket for oem retailers to offer correct/appropriate heatsinks. I believe some tests were done on some 130W yorkies and its was merely 100-110W on stock clocks, which is nice to know

Good point, oem companies demand accurate on-the-mark TDP values otherwise they will have to deal with hundreds or thousands of unstable systems. As you pointed out intel may just give those chips the worst case senario TDP values to be on the safe side when dealing with OEMs.

I am under the impression that the combination of 3.2 GHz and 1600 FSB is almost reaching the limits on how far YFs can be pushed on aircooling without using a top of the line heatsink in closed case conditions guaranteeing stability under full load or in otherwords still stay appealing to oems. Though intel may improve 9770s power consumption before release but the results I saw browsing through the net showed that the jump in power consumption going from 3GHz+1333FSB to 3.2+1600FSB was not in-line with how it was under 3GHz.

Here Anand thinks the QX9770 they had for preview testing adds a big chunk to the total system power consumption and is a bit of a hot CPU . I'd say it is, both literally and figuratively.

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3154&p=2
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3154&p=3

Intel probably knows every enthusiast worth their seed is going to get a badass cooler and overclock their chips to reach higher levels of performance. Nothing to worry, it's just a case of having politically correct TDPs for oems and server market imo.

 

[rhythmeister]

That is it, I am going Intel.
I have been running my Optiron 185 for over a year now waitting an answer from AMD for a quad core.
I want to support the little guy, but hey I need a new system for 2008

You NEED a new system? AM2 should suffice for a little while I reckon seeing as how it's cheap as chips, check out my sig rig for a gaming bargain

 

[rhythmeister]

Well pooh! I have been been staying away from the AM2's holding out for the AM2+'s. With this news and the current Phenoms and their L3 bug, the prices of the 690G boards and DDR2 800 are looking pretty affordable. I might just have to go the AM2 route (for now) with a Black Edition X2. Regardless, I won't be going to the blue camp.

With that rig you have you may as well bypass AM2 geezer, that TForce at 3.0GHz sounds like fun

 

[WarEagleAU]

Am I reading that right. The Tomalin (tri cored phenom) is a better price higher performing chip than intel quads? That surely cannot be right...