More conroe rumors...

[BrownTown]

So the Inq has some more Conroe rumors, these actually seems plausible, thoughts?

http://www.theinquirer.net/?article=29504

 

[BrownTown]

anyway, heres my thoughts:

- it says that this is directed at server vendors, so shouldn't that be woodcrest and not conroe?
- where is the 3.33G The Inq was pimping just a few weeks ago?
- also, where is the 1000$ conroe chip, top chip is only 530$, thatspretty cheap for a top of the line chip isnt it?
- 2.66 wont be enough to gain performance crown, it will probably split benchmarks with AMDs chips, not blow them out of the water like some are predicting

 

[Furen]

Yeah, I read this a few bit before, I just wanted it to be posted by someone other than me =)

Well, considering that the Inquirer mentioned a 3.33GHz, FSB1333 extreme edition part I was actually hoping that the regular parts would come out at, at least, 3GHz. It seems to me that either Intel is having trouble keeping these things as cool as it wants (these are the 50+W parts, after all) or Intel's extreme edition will really be extremely superior to the mainstream parts (I'd bet that if this is the clock at which the Exxxx CPUs are released the extreme wont be much over 3GHz, though). If those prices are right, however, it looks like the 2.67GHz part will be the top of the line part (or the second-best) and that doesn't sound too impressive considering that we're supposed to see a more mainstream 2.6GHz part when AM2 hits the market (I'd hardly consider the FX mainstream). Regardless I believe (or maybe hope?) that Intel is not clocking these as high as it can but rather leaving some headroom for clock-speed hikes if needed.

Now, about performance... I dont think a 2.67GHz part will blow AMD out of the water, I was thinking that a 3.33GHz would but unless Conroe is quite a bit more efficient than AMD's K8 (I dont see how it could be so but I've been surprised before, heh) then I think things will pretty much be tied... until AMD's 65nm shrink, I have no idea what kind of clocks AMD will be able to get from that considering that it only got about 400MHz from its 90nm shrink.

EDIT: I missed the "Server" thing... you are quite right, the FSB drop seems to make sense if these are the server parts, as does the lower clock. The prices dont make sense, however, I think they're too cheap to be server parts. Also, the model numbers seem to follow Intel's Yonah's (though at a higher power, hence the E instead of the T) but I havent seen anything that say what Intel will call its newer Xeons.

About the price. I think that's the $530 slot is the one that the D 950 currently occupies so the price doesn't seem to be mismatched.

 

[BrownTown]

my only guess is that this is only supposed to be the Q3 launch, so maybe they can push out 1 or 2 higher clocks before years end, but a 3.3G chip when the top of the line mainstream part is only 2.66 seems a pretty considerable reach... however, having said that, more then just the Inq have stated that a 3.3G chip does exist, so this confuses me, maybe Intel made them special just to send them around and ahveing everyone say that Conroe roxors, but still, thats 600+ mhz more

 

[Furen]

I find it quite unlikely. A 3.33GHz part would require a 12.5x multiplier on the same FSB compared to the 10 multi of the highest clocked E-series part, which is two speed grades and a half...

Did you notice that these seem to use whole multipliers on a 266MHz FSB? The next speed grade seems to be 2.93GHz which is quite a big jump, if you ask me. Sucks that the lower-end parts will be cache-halved parts so Intel is pulling an AMD in that regard...

 

[BrownTown]

well, the 3.3G is on a 1333 bus, so thats ony a 10x multi. But yeah, the 266mhz incriments are interesting since its means that there is a huge performance game between bins. How is AMD binning their CPUs \when they transition to 333mhz HT?

also, it doesn't say much about Intels process tech if AMD can have a 12 stage 90nm CPU at 2.8 and Intel can onyl get its 14 stage 65nm chip to 2.66, thats the main reason i thought that we would see higher speeds

 

[Furen]

I'm not sure. If AMD does indeed go to an HT of 333MHz then I'd guess it'll be in half multipliers. Tom's Hardware had an article with a pretty comprehensive list of CPUs to be available at AM2's launch and it didnt look like they'd be using a 333MHz HT, since the clocks were all in multiples of 200MHz. It shouldn't be too hard to do DDR2 667 even on a 200MHz HT since it would just mean that the memory would run around 20MHz slower than the spec (333MHz) on some parts... So really, who knows what the HT will end up being.

 

[BrownTown]

I'm pretty sure its 333

 

[Furen]

Hey, look at this: http://www.theinquirer.net/?article=29510
It doesnt make much sense for the server parts to be clocked so much higher than the desktop parts unless the server parts will launch quite a bit later. Maybe the inquirer has its clockspeeds messed up or something...

EDIT: Now that I think about it it actually does make sense. Intel said that Conroe would launch on a 1066 FSB (so that seems to be right on target) at a 65W TDP. Woodcrest was the part that was supposed to launch at 1333 FSB and it's also rated for a higher TDP, 80W.

 

[BrownTown]

my guess would be that woodcrest is coming out later than conroe, so they include the 3.0G part, and then they can launch the 2.93 desktop part late in '06, along with the 3.0G or 3.33 desktop EE

Of note: conroe TDP = 65W, woodcrest = 80W, Conroe XE = 95W, so conroe XE is SUPPOSED to clock higher than woodcrest when it comes out. So 3.3G is possible, but probably in late Q4, not July

EDIT: as far as performance goes, we know that the top of the line desktop part has to AT LEAST be able to easily beat a 3.6G Presler, becasue there is no way Intel would launch the 3.6G Presler and have it beating its new architecture sicne that would make for some really crappy PR (new Intel chip even worse then crappy old Intel chips)...

 

[Furen]

But damn, I hope Intel can really extract some clock speed from these CPUs through tweaks and stuff, it seems like it's already at the bleeding edge of the TDP.

 

[BrownTown]

i dunno, nowadays speed bumps just arent coming nearly as fast as they used too. I mean the P4 has been pretty well stagnent for like 2 years, the A64 is adding clockspeed slowly, and will certainly hit a limit soon at 90nm.

EDIT: this all seems to make pretty decent sense, Intel moved up the Conroe release date a couple of months, so they launch early at these speeds to try to get some leverage against AMD, then launch their 2.93 part around the same time that conroe was origionally scheduled to launch. The 2.66 conroe only has to deal with a 2.8G A64, and even a 2.66 Yonah could probably beat the 2.8 A64 is SuperPi (which means nothing, but some people will care). If conroe has 15% more IPC then Yonah (which is reasonable i think), then it will be able to match AMD, and at a cheaper price point too...

also, if this is something that many people are receiving then others should report it as well, THG is pretty much owned by Intel, why don't they post this if they have it?

EDIT2: now they say they have Merom numbers too... I wan't confirmation on all this, I mean anyone could pretty much have made these numbers up and be pretty clsoe the the truth, and if they are true others should report soon

 

[Cooler]

Originally posted by: BrownTown
i dunno, nowadays speed bumps just arent coming nearly as fast as they used too. I mean the P4 has been pretty well stagnent for like 2 years, the A64 is adding clockspeed slowly, and will certainly hit a limit soon at 90nm.

EDIT: this all seems to make pretty decent sense, Intel moved up the Conroe release date a couple of months, so they launch early at these speeds to try to get some leverage against AMD, then launch their 2.93 part around the same time that conroe was origionally scheduled to launch. The 2.66 conroe only has to deal with a 2.8G A64, and even a 2.66 Yonah could probably beat the 2.8 A64 is SuperPi (which means nothing, but some people will care). If conroe has 15% more IPC then Yonah (which is reasonable i think), then it will be able to match AMD, and at a cheaper price point too...

also, if this is something that many people are receiving then others should report it as well, THG is pretty much owned by Intel, why don't they post this if they have it?

that is logical for 4th quater
2.93 im gessing it will be at the ~$650 pirce point of the old 3.8 ghz P4s
3.33 EE of course this one will be $999

 

[BrownTown]

yeah Cooler, thats whay i'm guessing, Intel no longer has HT to justify EEs costing 1000$, so they need more clockspeed and a higher bus. So, if all goes as planed Intel should almsot certainly have the lead in Q4, then AMD sends in its 65nm chips and pretty much has to get it right or they will be down in performance until they can get 3.0G+ A64s out in the 65nm tech and in volume

 

[Brunnis]

I'm personally wondering a little about the power consumption of these chips. It doesn't seem to be very low. Sure, it's great compared to Prescott, but not so great compared to the current 90nm Athlon64. Is AMD going to win in the power consumption department with their 65nm products? Doesn't seem that crazy a theory...

 

[Diasper]

In a way I'd be expecting AMD to be starting speed binning for a 3ghz release on 90nm to meet Conroe's launch.

Anyone know when AMD are looking to introduce their new/latest SOI process tech? What process tech improvements are AMD planning to use and when?

As I see it AMD and Intel will be very equal when it comes to 'IPC' per clockspeed when AMD launches AM2.

- S939 & Yonah are about equal - Yonah perhaps a touch slower in some areas.
- AM2 AMD has a few core tweaks & increased FSB to benefit dual-core so we might see maybe 10% faster than S939 (AMD said 15%...) -while- Conroe has a more core tweaks and additional badnwidth meaning it might be 15-20% faster than Yonah for clock but since Yonah is a touch slower than S939 I'd expect there to be parity between Conroe and AM2 or rather at least within 10% performance of each other (in areas like games - encoding as ever I'm sure will be different).

Really, the competition between AMD and Intel is not really the design technology as they reach about parity (heck similiar design focus) atm but rather the process technology - who can fab smaller, cheaper, faster and with higher yields.

AMD and especially Intel are manufacturing companies first and design companies second. I wonder what ratio of R&D goes into manufacturing tech vs design tech - that'd make for some interesting analysis.

 

[Cooler]

Originally posted by: Brunnis
I'm personally wondering a little about the power consumption of these chips. It doesn't seem to be very low. Sure, it's great compared to Prescott, but not so great compared to the current 90nm Athlon64. Is AMD going to win in the power consumption department with their 65nm products? Doesn't seem that crazy a theory...

This might be the first EE worth geting just for the extream FSB.

 

[coldpower27]

Notice that the highest Server chip is Woodcrest 5160 3.0 with 1333FSB with 80W TDP, so I think an Extreme Edition @ 3GHZ with the same specs is possible at the 999US price point with 95W TDP.

Assuming Presler 960 is launched at the 637US price point, by the time Conroe ships it should be knocked down to 423US, with the Conroe E6700 2.66 taking the 530US pricing point, Conroe E6600 2.4GHZ will tie the Presler 940 at the 316US price point.

Then you have the Allendale E6400 2.13 vs the Presler 930 @ 241US and the Allendale E6300 1.86 vs the Presler 920 @ 209US.

 

[dmens]

also, it doesn't say much about Intels process tech if AMD can have a 12 stage 90nm CPU at 2.8 and Intel can onyl get its 14 stage 65nm chip to 2.66

maybe that's because merom probably is wider and has deeper buffers.

 

[Furen]

'm pretty sure Intel can get its 14 stage Conroe to at least 3GHz (the top-of-the-line woodcrest clock speed), the problem is getting it there while staying around the promised 65W power envelope. Kind of like how Yonah can probably hit 2.8GHz without much problems but your power consumption at that clock is probably hitting the 60+W range.

Also, remember that AMD has SOI, which is quite a bit better than bulk silicon on the static power side of things (I'm not too sure about dynamic power but I'd guess there's no benefit there) which gives it the ability to run at higher clock speeds as well. I'd say that SOI 90nm is kind of the mid point between bulk 90nm and bulk 65nm (with regard to static currents) but that's just a guess. Overall I think that the two manufacturing approaches are not very comparable.

 

[BrownTown]

so, looking at new roadmaps, it seems as though there are no plans to release a 2.93 Conroe until at least Q2 2007, however a Conroe XE is planned, which might be the 2.93 part. However, there are plans for Preslers at 3.6 and 3.73 before Conroes release. So are we supposed to think that at 2.66G Conroe can beat out a 3.73 Prelser?, becasue if not then that would mean that Netburst would be out performing NGMA, which wouldn't look to good for Conroe.

 

[coldpower27]

Conroe EE is rumored to be at least 3.0GHZ+ and with 1333FSB, aka like the highest Woodcrest, the rumors are pointing to a 3.33GHZ edition though interestingly enough.

What roadmaps indicate a 2.93GHZ/1066FSB Conroe at Q2 2007???

Also what roadmaps indicate 3.73GHZ Presler EE???

 

[Furen]

Rumors indicate that a 3.60GHz Presler will be coming out so I think he's infering that there will be a new EE will be launched as well...

 

[BrownTown]

no, this is on their roadmap...

roadmap

also interesting that they see the 3.6G Presler as the equal (at least in terms of price) of a 2.4G Conroe, how would a 2.4Ghz X2 compare to a 3.6G Presler?

of course it says nothing about the Conroe XE except that it will exist, so I guess the final specs are still undetermined (3.0 most likely, 3.3 if they are lucky)

 

[pedramrezai]

The way most of the people are thinking is just the way Intel has planned. I am a AMD fan but I can realize competition is good for customers. I was shocked by Core performance but using a handicapped AMD system really annoyed me. First, reviewers have already proven RD580 or solutions with dual 16x can deliver up to 10-15% more performance when paired with high-end, bandwith hungry vga cards. Second, we have been hearing of dual core optimizations in display drivers for some time but were unable to see something significant until we saw Conroe performance; I am quite suspicious over some hefty optimizations in intel-cooked display driver. Time will reveal. Third, this might be the beginning of a new SSEx game with unfair optimizations for a new technoogy.
I am surprized how people are trashing the current as well as future AMD64 technology.But remember that Core is not out yet and all these might be some optimizations that has granted it this performance level. Moreover, the current AMD64 technology is almost 3 years old and the new AM2 will update its specs. AMD did not like DDR2 high latency; What they are looking for is its higher frequency that can be paired with the new AM2 FSB.For Athlon 64 and Sempron a 333mhz FSB that paires with DDR2 666 and for the Fx parts a 400 mhz FSB pairing with DDR2 800. If DDR1 could reach these frequencies you could now see the real potential of AMD64. This kind of bandwith will give Core a hard time. Also remember that AMD is increasing cache (L2 and maybe L3). Shared cache is also something that will be seen in the future products and will bring huge performance gains. Based on the preliminary data of 200/266 async single channel bandwith of 3500mb/s a memory bandwith of >10k is expected in the final product and if Intel was going to compare its future platform, it was not fare to compare it with an infrastructure of >2 years old. I am sure the new AM2 will regain AMD reputation once again. But we all must remember that this competition between major players is good for the end users.