Sandy Bridge Numbers

[ydnas7]

My Conjecture on how Intel decides its model numbers

http://img251.imageshack.us/img251/2152/sandybridge.gif

 

[evolucion8]

Gotta love Intel's name scheme.

 

[ydnas7]

i just bashed it together from the other threads on this forum. I'ld say internally Intel organizes its products by bins/price point. with 2 dies, they have about 75 possible options for the start (and more to come each quarter), no wonder they will cull it down so not to confuse everybody!

 

[khon]

I think the price for the K-versions will be slightly lower than your estimate.

The 875K cost 40$ more than the 870, and the 655K is 30$ more than the 650. My guess is SB bridge models will have a pricing structure similar to that.

 

[Edrick]

If the Core i7-2600K is in the $300 price range, it should be a very good seller I would think. This chip should be able to hit 4+Ghz without much effort.

 

[ydnas7]

I think the price for the K-versions will be slightly lower than your estimate.

The 875K cost 40$ more than the 870, and the 655K is 30$ more than the 650. My guess is SB bridge models will have a pricing structure similar to that.

I revised the pricing guesstimate to reflect the above, its quite nice
http://img808.imageshack.us/img808/7233/sandybr.gif
Its possible (IMHO likely) that a 4C8T Sandybridge 'K' will always be cheaper than any Westmere whether 6C or 4C (2C turned off)

gulftown http://ark.intel.com/ProductCollection.aspx?codeName=29886

westmere http://ark.intel.com/ProductCollection.aspx?codeName=33174&code=Westmere-EP

 

[Fjodor2001]

Any idea what the price difference will be for the S and T versions of the SB CPUs (compared to the 'vanilla' version, i.e. the version without any specific letter)? Will they be cheaper or more expensive?

 

[khon]

Any idea what the price difference will be for the S and T versions of the SB CPUs (compared to the 'vanilla' version, i.e. the version without any specific letter)? Will they be cheaper or more expensive?

My guess would be more expensive.

Currently you can get an i5-750S and an i7-860S, both of which are $50-60 more expensive than the 'vanilla' versions.

The T-versions are new for SB, so the price for those is anyones guess, but I'd imagine they will be fairly expensive. If I were to take a guess I'd say ~$300 for the i5-2500T and ~$220 for the i5-2390T.

 

[gevorg]

Are these going to be available this year or next year?

 

[Fjodor2001]

Are these going to be available this year or next year?

Based on Intel's release schedule in recent years, they will be released at CES 2011 in January 2011. However, there have been rumours on the Internet saying that Intel may decide to release the Sandy Bridge CPUs earlier than normal. But at the moment anything is a guess. Perhaps Intel will provide more info on this at the Intel Developer Forum (IDF) in September.

 

[Fjodor2001]

My guess would be more expensive.

Currently you can get an i5-750S and an i7-860S, both of which are $50-60 more expensive than the 'vanilla' versions.

The T-versions are new for SB, so the price for those is anyones guess, but I'd imagine they will be fairly expensive. If I were to take a guess I'd say ~$300 for the i5-2500T and ~$220 for the i5-2390T.

Ok, I see. But isn't it weird though that the S and T versions are more expensive? Aren't they just 'vanilla' CPUs that didn't pass the tests checking if they function ok at the frequency that the vanilla chips use, so they had to be underclocked? Or do they in fact differ from the 'vanilla' CPUs in some other way, e.g. by being specifically tested to function ok when powered at a lower voltage (which the 'vanilla' CPUs do not function ok at)?

 

[LoneNinja]

Ok, I see. But isn't it weird though that the S and T versions are more expensive? Aren't they just 'vanilla' CPUs that didn't pass the tests checking if they function ok at the frequency that the vanilla chips use, so they had to be underclocked? Or do they in fact differ from the 'vanilla' CPUs in some other way, e.g. by being specifically tested to function ok when powered at a lower voltage (which the 'vanilla' CPUs do not function ok at)?

I may be wrong, but it's been my understanding that both AMD and Intels energy efficient line ups are nothing but the standard processors binned to be stable at lower voltages. They're probably also lower leakage parts.

 

[khon]

Or do they in fact differ from the 'vanilla' CPUs in some other way, e.g. by being specifically tested to function ok when powered at a lower voltage (which the 'vanilla' CPUs do not function ok at)?

That is my understanding of it.

The price is higher because these are parts that have been selected for superior performance at lower voltage.

 

[Fjodor2001]

I just found this interesting article on the Intel Core i5-750S CPU:

http://www.tomshardware.com/reviews/core-i5-750s-power,2557.html

Apparently, with the Core i5-750S the S variant of the CPU is simply underclocked to achieve the lower TPD.

Previously, with the S variant of the Intel Core 2 Quads (e.g. Q8200S and Q9550S), the S variant were hand picked CPUs that at a lower voltage were still stable at the same frequency as the non S variant of the CPU. Hence they did not have to be underclocked to achieve the lower TPD, and therefore had the same performance as the non S variant. I.e. the S variant of those Intel Core 2 Quads were true low power chips.

With Sandy Bridge the S variant of the CPUs run at a lower frequency than the non S variant. Therefore it seems like Intel has once again decided to take the cheap route (as with the Core i5-750S) and not provide hand picked low power CPUs for the S variant. This is in contrast to what the customer would expect, having paid a price premium over the non S variant of the CPU.

 

[khon]

With Sandy Bridge the S variant of the CPUs run at a lower frequency than the non S variant. Therefore it seems like Intel has once again decided to take the cheap route (as with the Core i5-750S) and not provide hand picked low power CPUs for the S variant. This is in contrast to what the customer would expect, having paid a price premium over the non S variant of the CPU.

Maybe, but I'm not so sure about that.

With the Lynnfield S-models the TDP only dropped from 95W to 82W, a mere 13W difference. With the SB models on the other hand TDP drops from 95W to 65W, a much more significant 30W difference. And that's even though the IGP still have the same clock rate, and thus presumably much the same power consumption.
To achieve that I think they'd have to pick lower power CPUs.

 

[gevorg]

Based on Intel's release schedule in recent years, they will be released at CES 2011 in January 2011. However, there have been rumours on the Internet saying that Intel may decide to release the Sandy Bridge CPUs earlier than normal. But at the moment anything is a guess. Perhaps Intel will provide more info on this at the Intel Developer Forum (IDF) in September.

Interesting. According to wiki though, Intel will release low and midrange in Q4 2010 and the highend/server class in Q3/Q4 2011.

 

[IntelUser2000]

Previously, with the S variant of the Intel Core 2 Quads (e.g. Q8200S and Q9550S), the S variant were hand picked CPUs that at a lower voltage were still stable at the same frequency as the non S variant of the CPU. Hence they did not have to be underclocked to achieve the lower TPD, and therefore had the same performance as the non S variant. I.e. the S variant of those Intel Core 2 Quads were true low power chips.

The S series on the Core 2 Quad uses the exact same voltage as the non-S series. The part is just better.

But its also way too simple to conclude the Lynnfield-based S is just crap. The Nehalem architecture adds significant amount of circuitry that's not related to the CPU, yet takes up die space and power.

Sandy Bridge would do better because they might have refined those circuits. It's a second generation chip for Intel that has integrated memory controller and QPI.

On the graphics chip side, if you look on mobile you can see the rate of progress has been diminishing quite rapidly in recent years as process variation and leakage becomes greater.

 

[ydnas7]

http://img714.imageshack.us/img714/7761/sandyr.gif

value comparison with current intel cpus