Intel Sandy Bridge Thread

[CAT-THE-FIFTH]

You have to remember Intel is primarily designing products for people who buy complete systems. There's no requirement to carry over ram, and designing a memory controller to cope with higher voltage than necessary creates additional cost and performance compromises. If the whole thing is meant to be energy efficient then why not lower the voltages where you can? It's a good move IMHO.

I don't agree with you at all as there is still loads of RAM on the market which runs at 1.6V. If anything this would lower the costs for system integrators.

Having THREE types of DDR3 RAM for Intel based systems in the last three years is excessive IMHO.

If anything it is the total shift to 32NM which is the main reason that the new processors have improved power consumption.

The voltage difference is tiny and smacks of BS. Like I said it only benefits the RAM companies who ATM are making very little money.

If power is so important then AMD with the AM3 Phenom II should have removed the DDR2 memory controller at the same time. On top of this they could have also forced lower voltage DDR3 RAM on their users as it would also help reduce the power consumption of their platform too. Considering that they are behind in process technology it would help.

Having two memory controllers actually cost them more money when it comes to die space and QC and QA(according to JF-AMD).

Yet the Phenom II processors worked with both the high voltage DDR3(used in the older Core2 systems) and lower voltage DDR3(used in the socket 1156) systems. AFAIK,they should work with 1.5V RAM too.

Like Intel their main markets is prebuilt computers too.

[kalniel]

I don't agree with you at all as there is still loads of RAM on the market which runs at 1.6V. If anything this would lower the costs for system integrators.

You can still use old stock if needed. My ancient corsair DDR3 stuff runs 1333 at 1.5v according to spd.

If anything it is the total shift to 32NM which is the main reason that the new processors have improved power consumption.

And as you drop to smaller processes you generally also can't handle higher voltages as easily.

If power is so important then AMD with the AM3 Phenom II should have removed the DDR2 memory controller at the same time. On top of this they could have also forced lower voltage DDR3 RAM on their users as it would also help reduce the power consumption of their platform too. Considering that they are behind in process technology it would help.

Having two memory controllers actually cost them more money when it comes to die space and QC and QA(according to JF-AMD).

Yet the Phenom II processors worked with both the high voltage DDR3(used in the older Core2 systems) and lower voltage DDR3(used in the socket 1156) systems. AFAIK,they should work with 1.5V RAM too.

Phenom II is 45nm right? That makes it easier for higher voltages, but in AMD's case compatibility is one of their competitive criteria, so it's worth the cost to add it in. Same reason Intel can include more robust controllers on their extreme chips. But there's no need for mainstream.

Like Intel their main markets is prebuilt computers too.

They don't seem to be as successful at it however..

[Terbinator]

As with Kaniel i don't buy the high-voltage DDR3 argument, 1.5v has always been the JEDEC standard AFAIK.

[CAT-THE-FIFTH]

You can still use old stock if needed. My ancient corsair DDR3 stuff runs 1333 at 1.5v according to spd.

If this is the case then it is good news but then negates the reason for these "1.45V to 1.5V" kits to come out at the same speeds.

And as you drop to smaller processes you generally also can't handle higher voltages as easily.

Phenom II is 45nm right? That makes it easier for higher voltages, but in AMD's case compatibility is one of their competitive criteria, so it's worth the cost to add it in. Same reason Intel can include more robust controllers on their extreme chips. But there's no need for mainstream.

The 45NM Core i5 processors could not handle above 1.65V but the 45NM Phenom II processors can.

They don't seem to be as successful at it however..

Which can be due to many factors though including the fact that Intel can spend much more advertising than a company like AMD could ever do.

Even when they were ahead with the Athlon 64 Intel still sold more processors AFAIK and more people knew about them.

They also abused their market position in the first place for years with companies like Dell.

On top of this AMD are capacity constrained(you should know this) as they are a small company and behind in process technology. Intel has been a world leader in fab technology for decades.

You do realise that fabs cost billions to run and develop and this was the main source of the losses for AMD. These means billions less on R and D.

AMD can sell every processor they made and Intel could still sell more. Intel can also expand capacity much more quickly too.

If anything Intel forced the situation on them so that they could not compete as effectively.

Hopefully things should improve though.

[CAT-THE-FIFTH]

As with Kaniel i don't buy the high-voltage DDR3 argument, 1.5v has always been the JEDEC standard AFAIK.

Making excuses for Intel are we??

Oh! Because Intel tells you it must be the way forward!

So why weren't the socket 1156 processors using 1.5V RAM ONLY??

Not even 1.55V but 1.5V RAM.

What about the fact that they changed sockets when they did not need to then??

Funny that socket 1156 processors will work with the new chipsets:

http://www.tomshardware.co.uk/p67-tr...iew-32092.html

[Terbinator]

Making excuses for Intel are we??

Oh! Because Intel tells you it must be the way forward!

So why weren't the socket 1156 processors using 1.5V RAM ONLY??

Not even 1.55V but 1.5V RAM.

What about the fact that they changed sockets when they did not need to then??

Funny that socket 1156 processors will work with the new chipsets:

http://www.tomshardware.co.uk/p67-tr...iew-32092.html

What!? 1.5v has been the standard since 2007.

Now if people wanted to buy borked kits of RAM that required more voltage than necessary then more fool them, going off Wikipedia 1.575v is the maximum recommended by JEDEC.
Of course, Intel must of told them to pick that number, no?

Again, the rampage about socket changes etc really has nothing to do with the RAM argument, which was my point.

[CAT-THE-FIFTH]

What!? 1.5v has been the standard since 2007.

Now if people wanted to buy borked kits of RAM that required more voltage than necessary then more fool them, going off Wikipedia 1.575v is the maximum recommended by JEDEC.
Of course, Intel must of told them to pick that number, no?

Again, the rampage about socket changes etc really has nothing to do with the RAM argument, which was my point.

Why did the DDR3 capable socket 775 motherboards work with high voltage DDR3 RAM then??

Then why were the socket 1156 processors validated to work with 1.5 to 1.65V RAM?

Why did Intel not force the 1.5V as the maximum voltage requirement for the last THREE years?

I honestly want to know.

[kalniel]

If this is the case then it is good news but then negates the reason for these "1.45V to 1.5V" kits to come out at the same speeds.

It negates the need, not the reason. The reason any new kits come out at the same speed is just marketing or lower B.O.M. Same way that the same chips can be sold in three different kits - high speed high latency high voltage, low speed low latency high voltage, and low speed high latency and low voltage.

The 45NM Core i5 processors could not handle above 1.65V but the 45NM Phenom II processors can.

Exactly - i5 didn't need to handle high voltage because it's a mainstream chip without any compatibility criteria. Note they also came out on one of the shortest livest sockets.

[Terbinator]

Why did the DDR3 capable socket 775 motherboards work with high voltage DDR3 RAM then??

Then why were the socket 1156 processors validated to work with 1.5 to 1.65V RAM?

Why did Intel not force the 1.5V as the maximum voltage requirement for the last THREE years?

I honestly want to know.

That was down to the NB then wasn't it?

Regardless, you still seem hung-up on this Intel= the standard setters. Yes, cynically they might be but as I've mentioned they're only going off the JEDEC spec, which was there in 2007 and was 1.5v (if not before?).

As for not limiting it to 1.5v, why would they? If they can support 1.65v voltages then thats their prerogative - not to mention the JEDEC recommended voltage maximum was 1.575v, not 1.5v.
Also, most 1.65v sanctioned/supported by Intel is by way of the XMP? The word extreme alone takes away any homogeneity in that regard.

[CAT-THE-FIFTH]

Firstly I need to clarify what the maximum voltage the new CPUs can handle.

I would assume 1.575V as this is the JEDEC specification(5% over 1.5V).

So,the new socket 1155 processors should handle 1.575V DDR3 RAM and not 1.5V DDR3 RAM as a maximum??

Wait a second!!

The Core i3 and Core i7 six core processors also could handle 1.5 V to 1.65V RAM AFAIK?? Even if you kept to the specifications this would mean a maximum of 1.575V!

They were 32NM too.

[Terbinator]

Firstly I need to clarify what the maximum voltage the new CPUs can handle.

I would assume 1.575V as this is the JEDEC specification(5% over 1.5V).

So,the new socket 1155 processors should handle 1.575V DDR3 RAM and not 1.5V DDR3 RAM as a maximum??

Wait a second!!

The Core i3 and Core i7 six core processors also could handle 1.5 V to 1.65V RAM AFAIK?? Even if you kept to the specifications this would mean a (JEDEC) maximum of 1.575V!

They were 32NM too.

Yes, but as i said if Intel think they can deliver at 1.65v theres nothing to stop them from doing so (does anyone actually know the RAM voltage 'limit' of SB because apparently vCore is 1.52 in Intels own papers, which seems high.)

[Champman99]

Who knew the voltage limits of RAM could create such firey debate!

Anyways - my SB rig is assembled. I ran it at stock for a little while just to see everything was working.

The Fenrir problem was solved by ... err.. buying a new TT Frio. Which I was a bit worried would block the Ram slots (the Vengenece memory have big heatsinks), but I'm OK as ASUS recommend using slots 2 and 4 anyway (if you have 2 sticks).

My 5850 also sits on one of the motherboards heatsinks rather awkwardly, but it was even worse in the other two slots (in the bottom pcie16x slot it blocked the front panel connections... and I do like to be able to you know, switch my PC on and off)


Right now I'm about to set Prime 95 running using 100mhz stock base clock, and a 44 Multiplier in turbo mode.

I'm pretty lost in the (non)BIOS tbh, I'm not completely sure that that will overclock all cores or whether just some, if all 4 are under load will it run all four at 4.4ghz? :S

Voltage I set manually to 1.25v, I'm not going to go beyond 1.3 at all. Too scared of all the bad reports.

I'll let you know if it passed the stress in the morning and put up some pics of the build if anyone is interested.



Edit: I'm a bit surprised at how warm its getting, not sure the Frio is on quite right, and also the Frios fans arent actually turned up full I dont think. Also the thermal compound needs to spread and set in.. I hope. Staying under 50, but I was expecting Icy cool!

[Terbinator]

With all four cores loaded you should still hit a multiplier higher than you've set in the BIOS. So for example, my 860 has a multiplier of x21(2.8GHz) but with all 4 cores loaded it goes up too 2.93Ghz - if there is thermal headroom.

[Champman99]

I turned the Frio fans right up just to be on the safe side. No failures overnight (all Prime 95 passed), peak temp was 56 before I turned the fans up and about 51 after.

All 4 cores were running at 4.4ghz under load (just checked it by changing the selected core in CPU-Z). So I'm very happy with that. Not going to push any further for now

[j.o.s.h.1408]

are there any i3 snady bridge? could not see one in scan.

im thinking of buying previous gen i3 and mobo's. far cheaper then a sandy bridge

[Terbinator]

are there any i3 snady bridge? could not see one in scan.

im thinking of buying previous gen i3 and mobo's. far cheaper then a sandy bridge

Release is 20th February apparently, according to Wikipedia.