What's AMD doing on sep. 12th?

[jones377]

Wasn't the previous LN2 overclocking record held by a single core P4 Prescott?

 

[BlueBlazer]

So thats over 1Ghz higher than a 990X in LN2, so can we assume it will OC just as much higher percentage wise on air?

The AMD overclockers are using LHe (liquid Helium) instead of only LN2 to push the limit. More over here....

Now AMD have only been testing the extreme overclocking capabilites of the FX processors for about a monthm but they were quickly able to determine that these new chips did not appear to have any cold bugs under liquid nitrogen, and this opened the door to something a little more extreme – liquid helium (Lhe). The hope was that by improving conductivity and decreasing operating temperatures to a few degrees above absolute zero, they might be able to achieve something amazing…and they did.

With a bunch of media in attendance, a few cherry-picked processors, and whole lot of liquid helium a supremely capable group made up of Sami (Macci) Mäkinen, Brian (Chew*) McLachlan, Aaron Schradin, Pete Hardman, and Simon Solotko pushed an octo-core FX-8150 (with two cores enabled) to new heights and achieved a new CPU frequency world record of 8429.38 MHz at 2.016V. This surpasses the current record by 119.44 MHz, a Intel Celeron 352 LGA775 chip clocked at 8308.9 MHz, and finally ends the streak of lowly “Cedar Mills” Celerons that have consistently held the title of world’s highest CPU frequency.

Anyone knows what SB does under LN2?

If not mistaken, around 6GHz.

 

[formulav8]

If not mistaken, around 6GHz.

Where did that number come from?

 

[grimpr]

They suffer from ColdBugs.

 

[LOL_Wut_Axel]

It looks like Bulldozer is AMD's new Pentium 4, only with more cores. If they can overclock high enough, though, they can be worth it, but I doubt we'll see more than 5.5GHz on high-end air cooling and at high voltage. Given that IPC is probably around the same, or a bit lower than Llano, it would mean that it'd need around 40% higher frequencies to match Sandy Bridge in single-threaded (because clock scaling decreases). In other words, Bulldozer would need around 6.3GHz to match Sandy Bridge running at 4.5GHz in single-threaded. Obviously, it has double the cores, so it should be at least a bit faster in multi-threaded.

 

[Borkil]

So I just got done talking to a guy from AMD and he said bulldozer should be out in beginning October. He also said that it would perform between a I5 and a I7. Unfortunately he didn't know about the announcement tonight.

 

[BlueBlazer]

Where did that number come from?

Example >> Core i7 2600K at 6003.4MHz.

They suffer from ColdBugs.

Yups.

So I just got done talking to a guy from Amd and he said bulldozer should be out in beginning October. He also said that it would preform between a I5 and a I7. Unfortunately he didn't know about the announcement tonight.

I hope it stays that way (and no more un-expected delays).

 

[Phynaz]

I doubt we'll see more than 5.5GHz on high-end air cooling and at high voltage.

Considering AMD got 4.8Ghz on water, I don't think we are getting anywhere near your estimate on air.

 

[nanaki333]

So I just got done talking to a guy from Amd and he said bulldozer should be out in beginning October. He also said that it would preform between a I5 and a I7. Unfortunately he didn't know about the announcement tonight.

i just got off the phone with an amd rep and he said that was just an ES that did that extreme OC. the retail version will be able to hit 8.8ghz on stock air cooling.

 

[Arkadrel]

It looks like Bulldozer is AMD's new Pentium 4, only with more cores. If they can overclock high enough, though, they can be worth it, but I doubt we'll see more than 5.5GHz on high-end air cooling and at high voltage. Given that IPC is probably around the same, or a bit lower than Llano, it would mean that it'd need around 40% higher frequencies to match Sandy Bridge in single-threaded (because clock scaling decreases). In other words, Bulldozer would need around 6.3GHz to match Sandy Bridge running at 4.5GHz in single-threaded. Obviously, it has double the cores, so it should be at least a bit faster in multi-threaded.

Llano is slightly slower than a Phenom II, a tiny bit.
JF-AMD said that the IPC increases compaired to the Phenom II.

The bulldozer mhz pr mhz, will be faster than a equal Phenom II, probably not by much but it wont be slower.

The fact that you can on air probably reach ~5ghz overclocking with these bulldozers is pretty good.

I realise you can do the same with a 2600k, and the 2600k will probably end up slightly faster in 4thread stuff, and the 8150 slightly faster with 8thread stuff.

 

[LOL_Wut_Axel]

Considering AMD got 4.8Ghz on water, I don't think we are getting anywhere near your estimate on air.

Since when is hybrid water cooling being considered "real water cooling"? It's not. It's worse than high-end air cooling.

Anyway, what I said is not an estimate. I said it'd probably be impossible to get anything higher than 5.5GHz on high voltage on air cooling.

 

[inf64]

So I just got done talking to a guy from Amd and he said bulldozer should be out in beginning October. He also said that it would preform between a I5 and a I7. Unfortunately he didn't know about the announcement tonight.

In many desktop workloads,Thuban already performs between i5 and i7... Does it make sense to introduce new design that does what previous one done before,only now maybe even worse?According to leaks sometimes 8150 is slower than 1100T...

 

[LOL_Wut_Axel]

Llano is slightly slower than a Phenom II, a tiny bit.
JF-AMD said that the IPC increases compaired to the Phenom II.

The bulldozer mhz pr mhz, will be faster than a equal Phenom II, probably not by much but it wont be slower.

The fact that you can on air probably reach ~5ghz overclocking with these bulldozers is pretty good.

I realise you can do the same with a 2600k, and the 2600k will probably end up slightly faster in 4thread stuff, and the 8150 slightly faster with 8thread stuff.

Llano has 2-3% higher IPC than Phenom II.

 

[Idontcare]

They suffer from ColdBugs.

My asus mobo has actually has an "LN2" switch on it, you flip the switch and it disables the cpu checks that cause cold-boot bug. It also disables a bunch of power-saving stuff so you don't really want to leave it enabled for 24x7 operation.

That said, there's no magical way to get around the 57x max multiplier. Once you hit 5.7GHz, doable on simple single-stage vaporphase, you are down to the mercy of OC'ing the BCLK.

No one really knows where SB the chip maxes out, we only know where the artificial limits are at.

 

[Borkil]

i just got off the phone with an amd rep and he said that was just an ES that did that extreme OC. the retail version will be able to hit 8.8ghz on stock air cooling.

lol yeah i guess that i don't sound very believable. anyways here's some creeper pics i snagged of the AMD booth where i heard that info

 

[BlueBlazer]

JF-AMD said that the IPC increases compaired to the Phenom II.

JF-AMD said a lot of things, but most are vague. I wouldn't put 100% trust on everything he said unless verified. :hmm:

The bulldozer mhz pr mhz, will be faster than a equal Phenom II, probably not by much but it wont be slower.

The fact that you can on air probably reach ~5ghz overclocking with these bulldozers is pretty good.

I realise you can do the same with a 2600k, and the 2600k will probably end up slightly faster in 4thread stuff, and the 8150 slightly faster with 8thread stuff.

Leaks are showing otherwise, Bulldozer can be slower than Phenom II (unless it can manage very high clock speed which means its IPC may be lower in desktop stuff). At same clock speeds IMHO I doubt that Bulldozer would be any faster than i7 2600K even with 8 threads. :hmm:

That said, there's no magical way to get around the 57x max multiplier. Once you hit 5.7GHz, doable on simple single-stage vaporphase, you are down to the mercy of OC'ing the BCLK.

No one really knows where SB the chip maxes out, we only know where the artificial limits are at.

Perhaps that's why i7 2700K may appear? The timing (of the leak) does seem coincidental.

 

[Edrick]

i just got off the phone with an amd rep and he said that was just an ES that did that extreme OC. the retail version will be able to hit 8.8ghz on stock air cooling.

Wait, what?

 

[Majic 7]

Wait, what?

Probably on all four modules. I believe we've been punked.

 

[LOL_Wut_Axel]

i just got off the phone with an amd rep and he said that was just an ES that did that extreme OC. the retail version will be able to hit 8.8ghz on stock air cooling.

Wrong.

10GHz on stock voltage on stock air cooling. 25GHz for the best samples, but on high voltage and water.















:sneaky:

 

[skipsneeky2]

Wasn't the previous LN2 overclocking record held by a single core P4 Prescott?

Think so that video by Toms hardware on youtube with the funky geeky music and them clocking it to 5.2ghzs i think?

Remember my p4 ocing days cheap 2.4ghzs 533 northy clocked it to 3.5ghzs on air.

Didn't replace it till the week the e6750 came out.

 

[WT]

I was actually going to purchase a 2500k tonight, and then read this thread. I don't honestly think I should change my decision, but at some point I am just tired of waiting and want to build a new rig.

I hope AMD's BD is a very capable chip and offers direct competition to Intel, but my problem is the waiting game. I'd surely be better off waiting another month just to get a comparison, but patience is not in my vocabulary.

 

[LOL_Wut_Axel]

I was actually going to purchase a 2500k tonight, and then read this thread. I don't honestly think I should change my decision, but at some point I am just tired of waiting and want to build a new rig.

I hope AMD's BD is a very capable chip and offers direct competition to Intel, but my problem is the waiting game. I'd surely be better off waiting another month just to get a comparison, but patience is not in my vocabulary.

What for? 90% gaming? If so, forget about Bulldozer already. If it's mostly for multi-threaded programs, then wait.

 

[ElFenix]

56 watt tdp according to the cpuz screen? :hmm:

 

[Majic 7]

56 watt tdp according to the cpuz screen? :hmm:

That's only one module. Maybe it's 224 with all four. BOOM

 

[Khato]

It looks like Bulldozer is AMD's new Pentium 4, only with more cores.

So true, with the key difference, and hence the reason why these overclocking results are boring indeed, is that all evidence available thus far says that the 'core' frequency they're reporting is purely for the integer logic. There is zero indication that the rest of the design runs at the same frequency as the integer core... which is, coincidentally, the same as was the case for the Pentium 4, except for the fact that Intel didn't advertise the integer core as being the processor frequency.

Now if AMD comes back and shows that all the functional logic is actually running at the indicated frequency, then it's actually somewhat impressive. If not, I believe that the fastest overclock of the Pentium 4's integer units was over 16GHz - it's comparatively easy to make the integer core run fast.