Rumour: Bulldozer 50% Faster than Core i7 and Phenom II.

[tijag]

If you have constrained production capability you make the chips with the highest profit margins first - that would be the bulldozer opterons. LLano is a low end consumer product and bound to have wafer thin margins hence to make much money you're going to have to sell a lot of them - not something you can do if you're production constrained.

An argument against that is that the Llano APU's represent AMD's future direction and market share which will allow them to push the OpenCL agenda and leverage their superior graphics capabilities might be more important long term.

If they gain some market share, then developer's will be interested in using the on die GPU for new and interesting things. Long term it might be more important to grab as much market share as possible with a consumer product, if that consumer product aligns with your current business plan.

 

[Nemesis 1]

Late summer, so august as a worse case scenario.

Ya right . August is best case . Sept is late summer. worse case scenario would be yet another stepping and 2012.

 

[Nemesis 1]

I can't quite make out what you're trying to say, but it seems you're implying that Pentium 4s outperformed Athlon 64s, which is simply untrue on every imaginable level.

Netburst didn't "destroy" K8 in any way. In fact, the high end (EE and 3.4+ GHz) Pentium 4s were the ONLY models to outperform K8 in audio encoding. Performance in encoding wasn't consistently in Intel's favor, and at the times it was, it was generally by under 5%. That said, when in AMD's favor, it was equally negligible.

Arguing that Netburst had anything on K8 is just as dumb as arguing that K10 had anything on Conroe.

No did not say that. But Until the X2 AMD couldn't win in a lot of benchies multi threaded . AMD x2 layed the wood on intel bigtime

 

[drizek]

AMD x2 layed the wood on intel bigtime

That's because Intel spent so much time laying pipe on Netburst.

 

[hamunaptra]

That's because Intel spent so much time laying pipe on Netburst.

LOL...

 

[Abwx]

An interesting clue...

"The order in which we launch those products, is of stategic nature only. The very pleasant success of C- and E-Series-APUs led to the decision to continue in the desktop-segment with APUs, for starters, and let the Bulldozer follow afterwards. Despite all rumours, there aren't any problems with Bulldozer." - Guido Lohmann, AMD PR Manager Northern Europe

 

[Vesku]

"PR Manager"

clue value = 1/0;

 

[Nemesis 1]

The truth will lie in GF 2nd Qt,report

 

[Abwx]

"PR Manager"

clue value = 1/0;

Yet, it has more value than your "argument" that consist
to denigrate the guy..

 

[Vesku]

Not denigrating the guy, just espousing the philosophy that PR statements are nearly as hard to categorize as any other bit of rumor or speculation. Of the cited statement I can only be mostly sure that AMD is pleased with Brazos.

 

[Abwx]

Not denigrating the guy, just espousing the philosophy that PR statements are nearly as hard to categorize as any other bit of rumor or speculation. Of the cited statement I can only be mostly sure that AMD is pleased with Brazos.

Since months we had infos suggesting that Llano was improving
and could stole the show to BD.

Since AMD is in short supply for Bobcat, it make sense to shift
priority to the most demanded products, keeping in mind that
Glofo 32nm capacities are still reduced , this automatically
imply far less wafers for BD.

Eventual BD wafers are probably reserved for servers Cpus since
they are to be used at lower frequencies that desktop variants,
and thus can be of the first batches, not counting a higher price.

 

[Vesku]

But the key is, as PR he wouldn't point out problems with BD regardless.

 

[Abwx]

But the key is, as PR he wouldn't point out problems with BD regardless.

Of course, he s not paid to perform auto bashings,
though, what he said make sense financialy speaking,
and we know that AMD has almost no cash..

Smaller chips provide better yields , not counting
that Llano is a high volume product while a DT BD
will never go beyond 10% of the total consumer products,
its sole lucrative market being, as already said, the server one.

 

[Abwx]

More on topic, a graph of the IPCs ,of some oldy Cpus , though,
but giving a good view about the rates that can be extracted
from the executions units..

 

[Dresdenboy]

More on topic, a graph of the IPCs ,of some oldy Cpus , though,
but giving a good view about the rates that can be extracted
from the executions units..

As the mcf result suggests, memory-related improvements like better prefetchers or memory disambiguation also play an important role. Execution units capabilities are just a part of the performance equation.

BTW the emulator MARSS is based on is PTLSim, which has been used by AMD people to simulate a K10-like arch.

 

[podspi]

"PR Manager"

clue value = 1/0;

Don't you mean 0/1?

More on topic, a graph of the IPCs ,of some oldy Cpus , though,
but giving a good view about the rates that can be extracted
from the executions units..

I've seen this before, would love to see it updated with some more modern Intel architectures like Nehalem and SB.

 

[RyanGreener]

Saw this on another forum:

Creator and Author of the highly successful CPU and Hardware Monitoring software CPU-z “Franck Delattre” has released the first official Bulldozer FX-Processor CPU-z Screenshot. The CPU-z screenshot details a FX-based processors (Engineering Sample) featuring 8 Cores and 8 Threads.

The new CPU-z 1.56.4 will add support for the FX-Series processors. Details in the screenshot show that the CPU is running at a clock speed of 1.4Ghz consuming 0.950V (TDP isn’t listed). Either the chip used was part of the older revision which had performance issues or due to power gating feature the cores were down clocked due to idle usage. Socket AM3+(942) has been mentioned and the chip has a total of 8mb L3 cache. The author also detailed some on the new features of Bulldozer CPU’s and added CPU-z support as follows:

- Support 2 Module 4-core (FX-4110), 3 Module 6-core (FX-6110) and 4 Module 8-core (FX-8110 8130P) Zambezi processor
– Support for Socket AM3 +
– 32-nm SOI HKMG Technology
– 213 million transistors per module
– Support HTT 3.1 bus (3.2GHz, 25.6GB / s, 16bit bi-directional transmission)
– Supports up to 8M L3
– Fundamental frequency of 3.2GHz (may be higher?)
– Turbo Core (up 1GHz, load up all the core 500MHz)
– Supports a maximum TDP 95W and 125W
– Support voltage VID 0.8V-1.3V
– Support for Advanced Power Management (Power Gating / Clock Gating / Sectored L3)

 

[Abwx]

Don't you mean 0/1?




I've seen this before, would love to see it updated with some more modern Intel architectures like Nehalem and SB.

According to this graph , a C2D run at an average of 1.2 IPC.
SB must be roughly at 1.6 IPC, so we can assume that BD s
two integer execution units are unlikely to be saturated and also
that they must be running at 80% of their max throughput
to equal SB s IPC.

We have infos that BD architecture is optimised to sustain
high IPC, that is, as constant as possible and close to
the max throughput.

Of course, things are different wether we take INTEGER of FP
to check the actual IPC.

 

[podspi]

According to this graph , a C2D run at an average of 1.2 IPC.
SB must be roughly at 1.6 IPC, so we can assume that BD s
two integer execution units are unlikely to be saturated and also
that they must be running at 80% of their max throughput
to equal SB s IPC.

We have infos that BD architecture is optimised to sustain
high IPC, that is, as constant as possible and close to
the max throughput.

Of course, things are different wether we take INTEGER of FP
to check the actual IPC.

Looking at Anand's benchmarks it does look like SB is around ~ 30% faster per clock than C2D. For some reason I expected it to be much faster. Still, I think it would be interesting to see some real-world results. It does look like ~ 2 IPC is that sweet spot between attainable and practical. We'll see how that works in practice... eventually.

 

[bryanW1995]

Not denigrating the guy, just espousing the philosophy that PR statements are nearly as hard to categorize as any other bit of rumor or speculation. Of the cited statement I can only be mostly sure that AMD is pleased with Brazos.

Generally speaking, they are actually worse. With random speculation you can at least assume that many/most posters have no agenda, but a PR guy is somebody paid to represent that comapany's interest. And often times the PR guy has no clue what he's talking about or really why something is delayed/changed/altered/cancelled/etc.

 

[AtenRa]

http://www.planet3dnow.de/vbulletin/showpost.php?p=4439215&postcount=2705

This is an ES sample


CPU-Z TXT Report
-------------------------------------------------------------------------

Binaries
-------------------------------------------------------------------------

CPU-Z version 1.56.4

Processors
-------------------------------------------------------------------------

Number of processors 1
Number of threads 8

APICs
-------------------------------------------------------------------------

Processor 0
-- Core 0
-- Thread 0 0
-- Core 1
-- Thread 0 1
-- Core 2
-- Thread 0 2
-- Core 3
-- Thread 0 3
-- Core 4
-- Thread 0 4
-- Core 5
-- Thread 0 5
-- Core 6
-- Thread 0 6
-- Core 7
-- Thread 0 7

Processors Information
-------------------------------------------------------------------------

Processor 1 ID = 0
Number of cores 8 (max 8)
Number of threads 8 (max 8)
Name AMD Processor
Codename Bulldozer
Specification AMD Eng Sample, 1D26246W8K44_36/26/22_2/8 (Engineering Sample)
Package Socket AM3+ (942)
CPUID F.1.0
Extended CPUID 15.1
Core Stepping
Technology 32 nm
TDP Limit 149 Watts
Core Speed 1400.0 MHz
Multiplier x FSB 7.0 x 200.0 MHz
Instructions sets MMX (+), SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, SSE4A, x86-64, AMD-V, AES, AVX, XOP
L1 Data cache 8 x 16 KBytes, 4-way set associative, 64-byte line size
L1 Instruction cache 4 x 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 4 x 2048 KBytes, 16-way set associative, 64-byte line size
L3 cache 8 MBytes, 64-way set associative, 64-byte line size
FID/VID Control yes
Min FID 7.0x
P-State FID 0x14 - VID 0x0B - IDD 12 (18.00x - 1.412 V)
P-State FID 0xE - VID 0x0E - IDD 10 (15.00x - 1.375 V)
P-State FID 0xA - VID 0x16 - IDD 10 (13.00x - 1.275 V)
P-State FID 0x7 - VID 0x1B - IDD 9 (11.50x - 1.212 V)
P-State FID 0x4 - VID 0x21 - IDD 8 (10.00x - 1.137 V)
P-State FID 0x1 - VID 0x26 - IDD 6 (8.50x - 1.075 V)
P-State FID 0x10C - VID 0x30 - IDD 6 (7.00x - 0.950 V)

Package Type 0x1
Model 00
String 1 0x0
String 2 0x0
Page 0x0
TDC Limit 96 Amps
Attached device PCI device at bus 0, device 24, function 0
Attached device PCI device at bus 0, device 24, function 1
Attached device PCI device at bus 0, device 24, function 2
Attached device PCI device at bus 0, device 24, function 3
Attached device PCI device at bus 0, device 24, function 4
Attached device PCI device at bus 0, device 24, function 5

 

[busydude]

Few more leaks..

3D Mark Vantage CPU Score:
AMD FX-8110(3.8Ghz): 81917 CPU Marks
Intel Core i7 2600K(Stock): 64146 CPU Marks
​

Looking fake on all accounts.. but this is a rumor thread..

 

[Ares1214]

Few more leaks..

Looking fake on all accounts.. but this is a rumor thread..

It probably is fake, but a score like that is very possible. BD looks like it will kill on synthetic benchmarks and server work, so it blowing SB away in this wouldnt surprise me. Whether or not it does the same in the real apps is what remains to be seen.

 

[Arachnotronic]

Uh huh, so a Bulldozer in 95W will get a CPU score that is >2x that of a 130W 990x in 3D Mark Vantage CPU score? *yawn*

Wake me up when something real leaks.

 

[psolord]

Guys, don't forget that Vantage is giving uber scores when Nvidia cards are used and PhysX is not assigned to cpu only in Nvidia control panel.

That being said, my Q9550@4Ghz+GTX 460@850Mhz (physx on), gave a cpu score of 48617. Now a GTX 560 Ti@900Mhz should be 20% faster than my 460, so the cpu score is still quite good, considering the aforementioned system, would barely manage to reach 55-60000 if it scaled linearly with the card's performance difference.

Heck even my i7+570 cannot go above 65000. So yeah, here's hoping this is true! We would be talking serious fpu power if it would be.