Propus Naked: New Die confirmed

[mAJORD]

Hey Guys..

This bastard was soldered!

Measurements are:

13.87mm x 12.37mm = 171.5mm2

As mentioned in the image below though, using my calipers, the dimensions come out slightly larger than official numbers on the other two chips.. So I'm thinking it's slightly smaller (SEE BELOW)






A Sneak peak at benchmarks under DDR3.






This very early silicon isn't the best I might add. This chip in particular has two issues:

1. A 1.4v Vcore.

2. Doesn't clock all that well compared to Regor.. topping out around 3.6Ghz under air.

Hopefully these chips will improve OR i have a poor example.

this high Vcore shows itself in Power consumption numbers:

Athlon II X2 250: 81w/117w
Phenom II 905e 2.5Ghz 65w: 84w/129w
Phenom II X2 550 (2core): 91w/138w
Athlon II X4 620: 86w/151w

This may disapoint HTPCers waiting for this chip, but have to see how they progress in later batches.

FURTHER RESULTS TO COME HOPEFULLY..

Anyway I'm going to Japan this week, so testing time is extremely stretched. I WANT to do a Deneb and Agena perhaps detailed Clock/Clock.. I've benchmarked Propus prior to removing IHS incase i wrecked it, but have to bench the others @ 2.6 also.

Anyway, Time to see if it still works!

[ubuntu83]

Geat job Waiting for the clock/clock results against Deneb.

[demonkevy666]

you got a leaker go put that thing under helium or neon >_>.....<_<........>_>........<_<.......
must be like

you still have IHS read write numbers off it my 940 here is 0047 last 4 digits in the second line. no matter what voltage on ait it's not going to see anything higher then 3.8ghz but under cold probably would be a good chip.

[justapost]

Thank you for the data.
Without L3 cache it is not possible to copy the L2 cache content to the L3 cache and power off the L2 when a core enters C1 state. This feature brought 2W lower consumption on denebs according to their electrical specifications.
Looking at the 250 results one core seems to require ~2.5W AC in idle. The specs say on the 905e four cores require 5W, adding vrm and psu efficiency ~10W for four cores seem reasonable. The only idle value which is abit off is your 550. It's electrical specifications in C1 are identical with the 905e.

http://support.amd.com/de/Processor_TechDocs/43375.pdf

[informal]

You did it man,never doubted you will try .Well done mAJORD!

Now let's see how stable the sucker is if you undervolt it to say 1.3V . Also,any tests for clock/clock comparison with Deneb and Agena are welcome

Good luck with the chip!

[imamage]

Hey Guys..

This bastard was soldered!

Measurements are:

13.87mm x 12.37mm = 171.5mm2

As mentioned in the image below though, using my calipers, the dimensions come out slightly larger than official numbers on the other two chips.. So I'm thinking it's slightly smaller (SEE BELOW)






A Sneak peak at benchmarks under DDR3.






This very early silicon isn't the best I might add. This chip in particular has two issues:

1. A 1.4v Vcore.

2. Doesn't clock all that well compared to Regor.. topping out around 3.6Ghz under air.

Hopefully these chips will improve OR i have a poor example.

this high Vcore shows itself in Power consumption numbers:

Athlon II X2 250: 81w/117w
Phenom II 905e 2.5Ghz 65w: 84w/129w
Phenom II X2 550 (2core): 91w/138w
Athlon II X4 620: 86w/151w

This may disapoint HTPCers waiting for this chip, but have to see how they progress in later batches.

FURTHER RESULTS TO COME HOPEFULLY..

Anyway I'm going to Japan this week, so testing time is extremely stretched. I WANT to do a Deneb and Agena perhaps detailed Clock/Clock.. I've benchmarked Propus prior to removing IHS incase i wrecked it, but have to bench the others @ 2.6 also.

Anyway, Time to see if it still works!

Thanks for sharing
Looks like I can keep my Phenom II x4 905e for my HTPC

[FischOderAal]

Thank you mAJORD

[Tom128]

That vCore seems really high for that speed. What board is it on and does it have official support for it? Maybe the auto-voltage is just not good? I know my 955BE auto-volted to 1.45v in my M3A79-T deluxe until 1103 bios (newest when I looked). I would say see how low you can get it to run at and then see the consumption!

[Peen]

My A2 X2 240 auto volted to 1.4v too but can undervolt to 1.08v at stock speeds 24hr prime

[Starscream]

grt pics.

2 days ago i picked up an Athlon X4 Propus.
Was kinda confused when i noticed it had 1.4vcore.
I actualy googled it to check if it was right cause i felt that it was to high.

I didnt have any time to play with it or to find out the lowest stable vcore but 1.3vcore worked perfectly fine and im shure it could have gone lower.

[mAJORD]

Thank you for the data.
Without L3 cache it is not possible to copy the L2 cache content to the L3 cache and power off the L2 when a core enters C1 state. This feature brought 2W lower consumption on denebs according to their electrical specifications.
Looking at the 250 results one core seems to require ~2.5W AC in idle. The specs say on the 905e four cores require 5W, adding vrm and psu efficiency ~10W for four cores seem reasonable. The only idle value which is abit off is your 550. It's electrical specifications in C1 are identical with the 905e.

http://support.amd.com/de/Processor_TechDocs/43375.pdf

That is interesting, and might even explain why the intersil VRM power meter (software) indicated higher idle power usage under CnQ on an Athlon II vs a Deneb :O

That vCore seems really high for that speed. What board is it on and does it have official support for it? Maybe the auto-voltage is just not good? I know my 955BE auto-volted to 1.45v in my M3A79-T deluxe until 1103 bios (newest when I looked). I would say see how low you can get it to run at and then see the consumption!

GA-MA770T-UD3P, F3 bios which is old but the newest available.

Will certainly check usage again if I can.

Informal: I'd say it'd be quite much lower at such a low clockspeed. It is stable up to around 3.35Ghz @ 1.35v


Cheers all.

[SocketMan]

Nice work many thx
Can you post the HSF pics - if possible?
I bet the A2X4 is dated 093x+


As far as the default voltage goes - it does depend on the particular CPU
even of the same model/stepping, as well as (imo) the motherboard.
Out of 3 (45nm) K10's that I have (710/955/x2 250) - all need 1.250v for
3.25ghz and 1.4v for 3.6ghz.

[FlanK3r]

good job,i have question, your Power comp. is with 1.4V at Athlon X4?

[justapost]

That is interesting, and might even explain why the intersil VRM power meter (software) indicated higher idle power usage under CnQ on an Athlon II vs a Deneb :O

Hmm after looking at the electrical specifications again it seems that feature is still present on athlon II's. I wonder where the L2 content goes.
They added the 250 to the specs. In C1 the 250's NB should require 1.7W and 905e's one 14.5W. Those specs make no sense to me looking at your results. I thought the IMC requires the biggest amount of power in idle and the L3 cache nearly zero.
You had C1E disabled for your tests? Here I gained a few watts by enabling dram power down channel based.

[chinook]

Excellent m8 how long did the IHS removal process take?
It's nice to see the HT clocked at 2GHz but what about the NB clock? same as HT at stock?

[Lightman]

Brill mAJORD!

I had 1.4Vcore on Athlon II 240, 245 and 250. It might be BIOS issue because at least my II 250 after flashing ASUS board got better and now is set to 1.35V from boot.

Other explanation is AMD artificially increasing it to get better yelds and still be way below 95W TDP

[Nedjo]

Brill mAJORD!

I had 1.4Vcore on Athlon II 240, 245 and 250. It might be BIOS issue because at least my II 250 after flashing ASUS board got better and now is set to 1.35V from boot.

Other explanation is AMD artificially increasing it to get better yelds and still be way below 95W TDP

it's definitely BIOS issue

quad Athlons are 95W parts, and no way def. voltage is 1.4V

[FlanK3r]

second review (first was at www.hardspell.com ):
Athlon X4 620
http://www.tweakpc.de/hardware/tests...x4_620/s02.php

[SocketMan]

Cmon AMD release the damn thing already (officially) so people can buy it here in Canada.

[G.Foyle]

one more day, my friend

[SocketMan]

Ordered !
Should be in store tomorrow:
http://ncix.com/products/index.php?s...anufacture=AMD

[mAJORD]

Well i'm off to japan in the morning, so before I go, here's the somewhat limited Deneb vs Propus clk/clk test.. Time to see the effect of L3 across the board.

I didn't have time to investigate why on earth Winrar yielded better results on propus (!) so i've got two total's, one omitting that result. When I get it'll be the first thing I check out.

Apart from that, test conditions are identical, no re-used data (Propus was tested first) , App's run a couple of times, and in the same order.



when I get back I hope to run a heap more and include gaming data, which should show a bigger advantage for Deneb I'm sure.

Cheers

[imamage]

Maybe the "Slow" L3 w/ Deneb slowing things down with WinRAR
Any chance to compare Propus with Agena (Phenom x4)
Thanks for sharing !

[Nedjo]

Maybe the "Slow" L3 w/ Deneb slowing things down with WinRAR
Any chance to compare Propus with Agena (Phenom x4)
Thanks for sharing !

WinRAR 3.9 x64 500 MB archiving:

Athlon II X4 620 - 154 sec
Phenom 9950 BE X4 - 162 sec

[Peen]

Newegg has in stock now for $99 Free shipping. So tempting!