New Llano leaks

[almighty15]

So by my reckoning BD is 15% faster then Denab?

Which make it's no faster then the old Nehealm based Core i7's.

It better be faster then that else I'm selling up and moving to Intel.

[jimbo75]

So by my reckoning BD is 15% faster then Denab?

Which make it's no faster then the old Nehealm based Core i7's.

It better be faster then that else I'm selling up and moving to Intel.

This isn't bulldozer, it's llano the midrange part.

[informal]

So by my reckoning BD is 15% faster then Denab?

Which make it's no faster then the old Nehealm based Core i7's.

It better be faster then that else I'm selling up and moving to Intel.

Just buy SB and be over with it.

[BeepBeep2]

This isn't bulldozer, it's llano the midrange part.

As far as I know, Llano and Bulldozer are spelled the same way

[Manicdan]

So by my reckoning BD is 15% faster then Denab?

absolutely right! more IPC, 2x as many cores, and faster clocks, but only 15% faster, lol

[FlanK3r]

Remember at first CPU-Z, BIOSes with Brazos platform? It showed 200 MHz baseclock (some Giga boards), later correct 100 MHz baseclock.
Im thnking, Llano is the same...User from coolaler see in BIOS 200 MHz, CPU-Z show 200 Mhz, but real is 100 MHz. Look at screen from HWINFO (default=100!):

[devguy]

I said Deneb and Deneb has L3. Propus has no L3,but I didn't use its numbers since it also has 2x less L2.

Yeah, take a look at the Regor dual-core Athlon II processor. It has 1mb L2 cache for each core and no L3 cache. Therefore, it performs closer to the Callisto Phenom II processor, than the Propus quad performs to Deneb.

[Mumak]

Damn, I forgot to resize the AMD VISION logo will fix that..
Nevertheless, the bus is 100 MHz (UMI) by default. Clocks should be fine too. Just the current multiplier plays crazy, I dunno why...
Notice that CPB (Core Performance Boost) is disabled there, so it can't go to turbo.
The Max clock (HFM) is in fact the CPU Max PLL clock.

[DarthShader]

Yeah, take a look at the Regor dual-core Athlon II processor. It has 1mb L2 cache for each core and no L3 cache. Therefore, it performs closer to the Callisto Phenom II processor, than the Propus quad performs to Deneb.

If I remember well, the benefit of the extra 512kb cache was estimated to be around 7%IPC.

[PatRaceTin]

i would like to see 3DMark series score

[Macadamia]

Proper screenshots at 3.7Ghz:
http://www.cbbimg.com/out.php/i81165...134DF0D34B.jpg
http://www.cbbimg.com/out.php/i81167...CD4DF0D8FC.jpg

so 3D06 is over 10k.

Also note how Easytune sees the CPU as a 3851.

[BeepBeep2]

Proper screenshots at 3.7Ghz:
http://www.cbbimg.com/out.php/i81165...134DF0D34B.jpg
http://www.cbbimg.com/out.php/i81167...CD4DF0D8FC.jpg

so 3D06 is over 10k.

Also note how Easytune sees the CPU as a 3851.

Dead links

[duploxxx]

The CPUMark score doesn't seem bad as my 4Ghz Thuban scores 553.

Liano score of the A8 is more or less equal to i5 2300.. just where AMD from a marketing perspective in there slides are rating it against
http://forums.overclockers.com.au/sh...d.php?t=954186

off course superpi scores are always way better on the intels...

Damn, I forgot to resize the AMD VISION logo will fix that..
Nevertheless, the bus is 100 MHz (UMI) by default. Clocks should be fine too. Just the current multiplier plays crazy, I dunno why...
Notice that CPB (Core Performance Boost) is disabled there, so it can't go to turbo.
The Max clock (HFM) is in fact the CPU Max PLL clock.

i think your CPB is not able to handle/read the values correct of the turbo 2.0, neither the GPU, it is somehow calculating from a wrong bus freq

[Macadamia]

Dead links

argh, I'm not rehosting because it's rude, so just go to the coolaler thread with the UD2H instead of the UD4H.

Page 3.
http://forum.coolaler.com/showthread.php?t=266864

[Mumak]

i think your CPB is not able to handle/read the values correct of the turbo 2.0

What? I say CPB is disabled on that machine. It might be because the machine is on battery power (on batt CPB is off)..
Wrong bus frequency is the result of invalid multiplier read. GPU clock should be ok.
I have already tested a different Llano model and there it worked ok..

[BeepBeep2]

5200 CPU score in '06...@ 3.8
Deneb @ 3.9 Ghz gets 5200.
Looks about the same as Deneb, maybe a tiny bit faster.

3D performance is about the same as that of an 8800/9800GT...I suppose you could call that impressive.

[SpuTnicK]

deleted.

[BeepBeep2]

1. This is a Llano thread
2. Repost of a screen we already found to be fake
3. The screen is fake
4. Screen was from a month ago
5. The screen was fake

[Mumak]

Also note, that there are (performance) errata in Llano B0, so I don't relay that much on performance results.

[jimbo75]

3dmark IGP world records tumble, unsurprisingly.

http://translate.googleusercontent.c...caQ#post396366

Is HWiNFO showing the Llano clocks correctly on coolaler?

[duploxxx]

interesting results, but the Q off course is what presets..... that A8 would run circles against its compettitor i5 GPU wise that is already something we knew just from a spec perspective.

GPU performance increase is huge when OC.

would like to see some CPU benches like cinebench and others to see how it fares against older Athlon and Phenom and the husky core enhancements.

Also note, that there are (performance) errata in Llano B0, so I don't relay that much on performance results.

what errata bug in B0 LIano ??? these are production SKU just like the link below. Are you mixing with BD?

[undone]

what errata bug in B0 LIano ??? these are production SKU just like the link below. Are you mixing with BD?

There was a frequency bug in Llano before, and I can sure it's B0 chips having bugs.
AFAIK I saw some screenshot before, some ES maybe B0 stepping, there's number 0 in the stepping blank. As for the benchmark on that SKU, it's also 0 stepping, but I can't sure if it's a B0 chip, maybe it's C0/D0/E0 & so on. If that SKU is B0 it could have some bugs.

[duploxxx]

both Coolaler and Radeonfanboy have retail SKU -> B0 stepping, sure you are not confusing with BD ????

CPU-z wasn't able to read the correct Bus speed and not able to translate the turbo that is all

[undone]

both Coolaler and Radeonfanboy have retail SKU -> B0 stepping, sure you are not confusing with BD ????

CPU-z wasn't able to read the correct Bus speed and not able to translate the turbo that is all

No I'm not confusing with bulldozer. Was Coolaler find the frequency bug of Llano before? It was a B0 chip. Soon clarified by those taiwanese, that system couldn't read the frequency correctly. So I guess that was a ES chip but not a final retail one. As for the llano chip that tweaktown have, it could be buggy as well, of course it could also be a healthy retail one too.

[Mumak]

No, I'm not mixing anything.