AMD Zambezi news, info, fans !

[XRL8]

Well ,maybe B0 ES`s have this abnormally high voltage, sans 186W max TDP rating.Cpuz creator has one.Its not weird for ES to have high leakage.

[repman]

The 186W TDP is CPUZ error I think, look at HWiNFO, it says ~125W

[demonkevy666]

The 186W TDP is CPUZ error I think, look at HWiNFO, it says ~125W

new one supports bulldozer now 1.58

[duron]

True? No?

An overclocker from the Czech Republic has managed to obtain an engineering sample of AMD FX-8130P microprocessor code-named Zambezi that is powered by Bulldozer micro-architecture and overclock it to 4.6GHz.

http://www.xbitlabs.com/news/cpu/dis...to_4_6GHz.html

[rintamarotta]

What you would say if i would say Zambezi ES can do 4.6Ghz with 1.4volts wPrime 1024M stable?

[BeepBeep2]

What you would say if i would say Zambezi ES can do 4.6Ghz with 1.4volts wPrime 1024M stable?

IF true, I would be pleasantly suprised.

I want these chips to do at least 4.6-4.8 24/7...5 Ghz while a stretch I'd like to see these chips do a year from now.

[FlanK3r]

wow! It can be very good. 4.6 GHz with 1.4V is not bad. Oh, guys, I have new one CPU ,-)...A bit oldschool, A3200+ s939 , my first 939 CPU

[zalbard]

You can not judge whether voltage is too high or low. This is a new architecture. Wattage is what matters.

[TheBreezyBB]

I want BD to reach a 4.8GHz OC @ 1.4

[Olivon]

What you would say if i would say Zambezi ES can do 4.6Ghz with 1.4volts wPrime 1024M stable?

I would say : Pics or didn't happened

[Manicdan]

not sure if i would care about the ES results, or the voltage needed, cause who knows what the retail samples will do

for me to be happy with BD, i want 5ghz done for 90% of us on air, and a little higher for water.
and it will be really interesting to see how well 4 vs 8 cores work with gaming due to sharing modules.

[Tomasis]

5ghz cinebench stable for retail would be awesome! there is room to improve from ES

[Lightman]

Aren't you hoping for too much?
Once core 5GHz on AIR would be very good achievement!
8 cores at 4.6GHz is simply amazing. Just project CB11 score using Thuban with 2 extra cores at that clock It should be better, but lets aim low!

[BeepBeep2]

I predict a pretty hefty IPC improvement here...
The Llano stars cores are ~5% improved over Propus, and if you threw all the cache of Phenom II plus some, made that cache faster, you'd probably be looking at a good 5-10% over Deneb.

That SHOULD mean easy guaranteed 10-20% IPC increase over Phenom II IMO...but we will see.

[Particle]

for me to be happy with BD, i want 5ghz done for 90% of us on air, and a little higher for water

You aren't being very reasonable. You're almost certain to be disappointed.

[BeepBeep2]

You aren't being very reasonable. You're almost certain to be disappointed.

ES = fab cold
Retail = hopefully fab hotter

We will see if he is dissapointed or not.

Considering a simple Stars shrink to 32nm should do 4.6-4.8 Ghz I would guess (...forget Llano APU, though it is clocking well), I'd expect either 4.6-4.8 with 15-20% IPC improvement, or near 5.0 with not such a large IPC improvement.

[Manicdan]

You aren't being very reasonable. You're almost certain to be disappointed.

if i have ~200$ to spend on a gaming cpu, after i overclock, which will be better a 2500k or 4/6 core BD?
the 6 core BD will have better IPC than the 4 core, and i wont really be concerned with TDP, so if they reach the same clocks, i expect the 4core to loose
but compared to the IPC and overclocking of SB, its tough to know how well BD will stack up compared to it. if i wanted 8 threads max perf, i expect BD to win obviously, but i care most about 3-4 threads

[FlanK3r]

guys, relax....it is not possible at first SOI HK procces get 5 GHz aircooled Cinebench, thinking a bit. I think, real values are between 4.5-4.7 GHz.

[richierich]

Well it seems that most 2500Ks can hit 4.5GHz pretty easily. BD cores would need to hit 4.5GHz+ to match it. Cmon AMD, you can do it!!!

[FlanK3r]

bud in multithreading has no change 2500k against 4-module BD....Remember...32nm first proces was about 4200-4500 MHz stable at Gulftown in 2010!

[BeepBeep2]

Air -
Thuban - A lowly 4 to 4.2 Ghz
Bulldozer - ????
Bloomfield - upwards of 4.2 Ghz
Lynnfield - Same
Deneb - Seem to be capable of doing a good 4.2-4.3 now, none of us buy new chips to test
Sandy Bridge - 4.6 to 4.8 Ghz

Water
Thuban - 4.1 to 4.4 Ghz
Bulldozer - ????
Bloomfield - upwards of 4.4 Ghz
Lynnfield - about same
Deneb - seem to be capable of 4.3-4.4
Sandy Bridge - say hello to 5 Ghz.

From Deneb, AMD needs at LEAST 20% IPC improvement to match Bloomfield...which in my opinion would make a lot of us happy.

Seeing how Llano performs with such little cache it has, I'd say were looking at 10-15% if they stuck with that architecture and added improvements with L2/L3 cache.

So, how do we find five more percent?

[XRL8]

You guys seem to forget that this BD cores are neither K10 based, nor a real cores.
This simple math just isnt going to cut it, the cores itself are much smaller than the SB ones, so dont hope that theyre gonna be in the same league.
From what i understood however ,amd will gonna try to boost their simple cores in light workload to achieve some parity.
Comparing BD cores to stars cores isnt really sensible too ;-) .Its a new architecture.
It is supposed to be high clocking one too.So i would expect 5ghz on high end air, at least if its going to win with SB.
Most probably ,depending on the workload, youre gonna see a BIG boost, and in another cases none.Thats probably the thing with superpi.
On the other spectrum you have truecrypt which uses new instructions , so its gonna have HUGE boost.

[BeepBeep2]

...
If IPC were to be worse than Deneb then Stars X8 shrunk with fast/loads of cache would be the answer.
Based on the cores in Llano they could potentially sell 10 core desktop chips with that design.

Shrunk stars (apparently the core gains ~5%) + 5% in cache improvements and look what you have, 10% IPC improvement. Llano has hit 4.6 already, and it has a f(l)ucking GPU on the die, so we could have possibly seen an 8 core based on stars cores doing 4.6 to 4.8 Ghz.

Obviously one would believe that Bulldozer is better.

[tbone8ty]

Ill be happy when they finally release it and i can order it off of newegg

[Motiv]

You guys seem to forget that this BD cores are neither K10 based, nor a real cores.
This simple math just isnt going to cut it, the cores itself are much smaller than the SB ones, so dont hope that theyre gonna be in the same league.
From what i understood however ,amd will gonna try to boost their simple cores in light workload to achieve some parity.
Comparing BD cores to stars cores isnt really sensible too ;-) .Its a new architecture.
It is supposed to be high clocking one too.So i would expect 5ghz on high end air, at least if its going to win with SB.
Most probably ,depending on the workload, youre gonna see a BIG boost, and in another cases none.Thats probably the thing with superpi.
On the other spectrum you have truecrypt which uses new instructions , so its gonna have HUGE boost.

If you have the 8 core version (4 module), it will beat intels chips when 8 cores are utilised. If people want single thread performance, then Intel will undoubtedly have that area sawn up. If you're a gamer, then a 4 core intel will be a beter chip than a bulldozer.

It's no wonder AMD are going for the server market, as these chips will sell like off a shovel in the multi-threaded environment. Desktop wise, just buy what you can afford, unless you want the best, then get an intel.