Sneak peak at FX-8150 with SuperPi & wPrime

[BlueBlazer]

54 Seconds on a Pentium D 820 @2.8 GHZ LMAO

Mine....

Pentium4 1.7GHz >> 107 seconds
Athlon XP 2000+ >> 77 seconds

Even if I've slightly overclock that Pentium4, it will not catch up...

 

[BlueBlazer]

I just explained this. For the same reason why you shouldn't compare K10.5 and Sandy Bridge using DiRT 3. There's applications that are made to take advantage of some CPUs over the others, so they shouldn't be the main ones we look at. Instead, we should discard them for the most part as an overall outlook on performance.

*snip*

The benchmark was made to take advantage of K10.5 and Nehalem yet not of Sandy Bridge. Should we use it to base performance, therefore? No.

That program was optimized for certain CPU architectures. However SuperPi is not optimized at all for any CPU architecture.

wPrime isn't very good, either. It puts the Atom D510 as having similar performance to a Core 2 Duo SU7300.

wPrime is a multi-threaded benchmark, and the Atom D510 has 2 cores/4 threads versus Core 2 Duo SU7300's 2 cores/2 threads.

 

[Madcatatlas]

I still wonder why they even bother to post SuperPi results. It uses the archaic X87 instruction set and is not good at all for disseminating single-threaded performance. It's bad for modern CPUs, even more for AMD's since it gives Intel's architecture a huge advantage.

Just to point out how bad SuperPi is, the old CPU in my laptop (Core 2 Duo P7450, now a T9600) overclocked to 2.7GHz calculates 1M in 19.465s, beating the FX-8150 by more than a whole second while being clocked 900MHz lower.

And yes, it was using SuperPi mod 1.5.

ooh, my CPU did it in 10.777

is this good?

 

[T_Yamamoto]

im gonna try it with my brothers laptop

 

[T_Yamamoto]

15.225 with my brothers i5 2410 @2.3 GHZ

 

[TakeNoPrisoners]

My x4 955 at 3.8 GHz does Super Pi mod 1.5 does 1m in 17 seconds. This is unimpressive.

 

[drizek]

sigh...

 

[T_Yamamoto]

wow. my Pentium is faster

 

[LOL_Wut_Axel]

That program was optimized for certain CPU architectures. However SuperPi is not optimized at all for any CPU architecture.

wPrime is a multi-threaded benchmark, and the Atom D510 has 2 cores/4 threads versus Core 2 Duo SU7300's 2 cores/2 threads.

As has been presented to you time and time again, SuperPi heavily favors Conroe over K10.5. Therefore, it shouldn't be used to judge overall performance.

I'm not sure if you're trolling, but I'm thinking you're reaching that borderline level.

Atom has slightly LOWER IPC than Pentium 4, and Penryn has 110% higher IPC than Pentium 4. Therefore, Penryn has more than 110% higher IPC than Atom. The Atom 510 has HyperThreading, but even in the heaviest multi-threaded benchmarks that's a max increase of around 25%. The Atom is also clocked 23% higher. So let's take the best-case scenario and add those numbers and you still have at least 60% performance that has gone "unaccounted for". Care to explain that?

 

[Abwx]

wPrime is a multi-threaded benchmark, and the Atom D510 has 2 cores/4 threads versus Core 2 Duo SU7300's 2 cores/2 threads.

True , but not very well...
Indeed, the results are surprising...

Seems that the added hardware divider and multiplier in Llano s
ALUs are doing miracles...

 

[StinkyPinky]

Don't panic guys. Apparently this CPU will be running at 1.21 gigawatz in its final silicon. Raw clock speed is the key.

 

[BlueBlazer]

As has been presented to you time and time again, SuperPi heavily favors Conroe over K10.5. Therefore, it shouldn't be used to judge overall performance.

I'm not sure if you're trolling, but I'm thinking you're reaching that borderline level.

During the days when SuperPi has been shown to favor Athlons then this is an essential benchmark (one that to shows K7 and K8's advantages). But nowadays when SuperPi seemingly "favors" Intel architecture, then it is no longer a good benchmark. I guess wPrime (and possibly Cinebench) will be next (to be dismissed)?

Atom has slightly LOWER IPC than Pentium 4, and Penryn has 110% higher IPC than Pentium 4. Therefore, Penryn has more than 110% higher IPC than Atom. The Atom 510 has HyperThreading, but even in the heaviest multi-threaded benchmarks that's a max increase of around 25%. The Atom is also clocked 23% higher. So let's take the best-case scenario and add those numbers and you still have at least 60% performance that has gone "unaccounted for". Care to explain that?

The Atom D510 is not based on Pentium4 architecture, rather its based on Pentium-M architecture (the precursor to Core Duo and Core 2 Duo). Thus your assumptions are incorrect. You can check this Intel Pentium 4 Vs. Atom: A Battle Of The Generations, the Atom D510 can sometimes outperform Pentium4 depending on applications used.

 

[Arkaign]

During the days when SuperPi has been shown to favor SuperPi

I love it when SuperPi favors SuperPi, it's awesomely redundant

 

[BlueBlazer]

I love it when SuperPi favors SuperPi, it's awesomely redundant

Oops, corrected. Thanks for heads up.

 

[LOL_Wut_Axel]

During the days when SuperPi has been shown to favor SuperPi then this is an essential benchmark (one that to shows K7 and K8's advantages). But nowadays when SuperPi seemingly "favors" Intel architecture, then it is no longer a good benchmark. I guess wPrime (and possibly Cinebench) will be next (to be dismissed)?

The Atom D510 is not based on Pentium4 architecture, rather its based on Pentium-M architecture (the precursor to Core Duo and Core 2 Duo). Thus your assumptions are incorrect.

You've GOT to be trolling. No one on a tech enthusiast forum is this clueless.

I NEVER said Atom was based on the Pentium 4 architecture. I said it has LOWER IPC than Pentium 4, which is a well-known fact. Atom is based on the Bonnell architecture, not Pentium M like Yonah (Intel Core).

It has lower IPC than Pentium 4 because it uses in-order execution and a 2-issue design.

Also, NO ONE here said that Cinebench is not good for comparing multi-threaded performance because it is. It's largely CPU-agnostic and gives an accurate representation. If Bulldozer is scoring so poorly in it it's because of either a bug or a design issue, or IPC is significantly lower than K10.5 (which I doubt).

As for your last sentence, Atom loses against Pentium 4 in single-threaded. Only its two cores plus HyperThreading save it in the article you linked.

Also, Core 2 Duo SU7300 vs Atom D510 in wPrime 32M:

SU7300

62 seconds

Atom D510:

61.8 seconds.

There's no way an Atom D510 is even close to the performance of a Core 2 Duo SU7300. None.

 

[Nemesis 1]

No matter how ya see it . These benchies with superPi is alot better than that phony AMD benchtest were AMD claims victory

 

[LOL_Wut_Axel]

No matter how ya see it . These benchies with superPi is alot better than that phony AMD benchtest were AMD claims victory

Nah, they're just as bad. SuperPi is too pessimistic, DiRT 3 seems like a best-case scenario looking at the fact that's what AMD is using and it performs so well on the Phenom II X6, even better than the Core i5-2500K in fact.

 

[BlueBlazer]

You've GOT to be trolling. No one on a tech enthusiast forum is this clueless.

I NEVER said Atom was based on the Pentium 4 architecture. I said it has LOWER IPC than Pentium 4, which is a well-known fact. Atom is based on the Bonnell architecture, not Pentium M like Yonah (Intel Core).

It has lower IPC than Pentium 4 because it uses in-order execution and a 2-issue design.

The original Atom was based/derived on the Pentium-M architecture, without the Out-Of-Order unit and a few other stuff (to further reduce power consumption). Bonnel was the new codename for this new architecture.

Also, NO ONE here said that Cinebench is not good for comparing multi-threaded performance because it is. It's largely CPU-agnostic and gives an accurate representation. If Bulldozer is scoring so poorly in it it's because of either a bug or a design issue, or IPC is significantly lower than K10.5 (which I doubt).

Need more leaks (quite a number already showing more or less similar performance), or wait for the official benchmarks (unknown since AMD has been so "quiet") to confirm. :sneaky:

As for your last sentence, Atom loses against Pentium 4 in single-threaded. Only its two cores plus HyperThreading save it in the article you linked.

Look again, do contemplate on clock speed differences between the two (the Pentium4 is a Northwood, not Prescott and clocking at nearly 2X the frequency of the Atom D510).

 

[Puppies04]

8.670 secs on my 2500k at 4.4ghz

 

[LOL_Wut_Axel]

The original Atom was based/derived on the Pentium-M architecture, without the Out-Of-Order unit and a few other stuff (to further reduce power consumption). Bonnel was the new codename for this new architecture.

Need more leaks (quite a number already showing more or less similar performance), or wait for the official benchmarks (unknown since AMD has been so "quiet") to confirm. :sneaky:

Look again, do contemplate on clock speed differences between the two (the Pentium4 is a Northwood, not Prescott and clocking at nearly 2X the frequency of the Atom D510).

Atom IS NOT based on the Pentium M architecture. Just the large list of performance-reducing and power-reducing "tweaks" they did completely cripple it. The fact it uses in-order execution alone makes a HUGE difference, and because of that alone you can say it's not based on it.

It's so crippled that it has slightly lower IPC than a Pentium 4, and that's one of the worst chips to have come out in that regard. The fact that even taking this into account it can be faster than a CULV Core 2 Duo in wPrime is baffling.

I'm not gonna feed you anymore. Go troll somewhere else, goodbye.

 

[TakeNoPrisoners]

Based on my SuperPI results and these benchmarks I think the single core performance per clock is the same as PII. If the Bulldozer quad core can overclock past 4.5GHz to maybe 5GHz then this might be a good upgrade from X4 955.

 

[BlueBlazer]

Atom IS NOT based on the Pentium M architecture. Just the large list of performance-reducing and power-reducing "tweaks" they did completely cripple it. The fact it uses in-order execution alone makes a HUGE difference, and because of that alone you can say it's not based on it.

It's so crippled that it has slightly lower IPC than a Pentium 4, and that's one of the worst chips to have come out in that regard. The fact that even taking this into account it can be faster than a CULV Core 2 Duo in wPrime is baffling.

Even though it was crippled, the original architecture from which the Atom was derived is highly efficient. It does not necessarily have a lower IPC than Pentium4 most of the time. From here, for example take a look at Fritz Chess (synthethic, multi-threaded)....

Atom D510 1.66GHz >> (1597 / 4 threads) / 1.66GHz = 240.5
Pentium4 3.2GHz >> (1148 / 2 threads) / 3.2GHz = 179.375

In Fritz Chess application, the Atom D510 is 34% faster per thread and per GHz clock, thus per core (2 threads) Atom D510 is 68% faster, and for 2 cores (4 threads) Atom D510 is 136% faster than Pentium4 at the same clock speed.

I'm not gonna feed you anymore. Go troll somewhere else, goodbye.

I've just shown that Atom D510 actually can have higher IPC than Pentium4. Guess you couldn't stand the truth. :hmm:

 

[blackened23]

My 486 just got 70 fps in doom -timedemo, how does BD do?

 

[Nemesis 1]

Even though it was crippled, the original architecture from which the Atom was derived is highly efficient. It does not necessarily have a lower IPC than Pentium4 most of the time. From here, for example take a look at Fritz Chess (synthethic, multi-threaded)....

Atom D510 1.66GHz >> (1597 / 4 threads) / 1.66GHz = 240.5
Pentium4 3.2GHz >> (1148 / 2 threads) / 3.2GHz = 179.375

In Fritz Chess application, the Atom D510 is 34% faster per thread and per GHz clock, thus per core (2 threads) Atom D510 is 68% faster, and for 2 cores (4 threads) Atom D510 is 136% faster than Pentium4 at the same clock speed.

I've just shown that Atom D510 actually can have higher IPC than Pentium4. Guess you couldn't stand the truth. :hmm:

Nice job of de masking . . . Your doing a really fine job . Don't let guys like these interferr with your research. Keep at it. Hammer away at these threads.

 

[LOL_Wut_Axel]

Even though it was crippled, the original architecture from which the Atom was derived is highly efficient. It does not necessarily have a lower IPC than Pentium4 most of the time. From here, for example take a look at Fritz Chess (synthethic, multi-threaded)....

Atom D510 1.66GHz >> (1597 / 4 threads) / 1.66GHz = 240.5
Pentium4 3.2GHz >> (1148 / 2 threads) / 3.2GHz = 179.375

In Fritz Chess application, the Atom D510 is 34% faster per thread and per GHz clock, thus per core (2 threads) Atom D510 is 68% faster, and for 2 cores (4 threads) Atom D510 is 136% faster than Pentium4 at the same clock speed.

I've just shown that Atom D510 actually can have higher IPC than Pentium4. Guess you couldn't stand the truth. :hmm:

You're on ignore, but Nemesis quoted you. Some things: you compare using cores, not threads. If you want to see IPC, you use a single-threaded application, preferably at the same or close clock speed.

The implementation of HyperThreading in the Pentium 4 is very poor and at many times actually hinders performance, while on an Atom it can bring up to a 20% improvement. That alone makes your comparison invalid.

And yes, Atom does have lower IPC than Pentium 4. That's been known for a long time now, and the main reason why is that even though it doesn't have a deep pipeline, it has in-order execution. Intel's focus with Atom wasn't energy efficiency, but providing "decent-enough" performance and very cheap manufacturing costs. If you compare the energy efficiency of Atom to a Core i3, it'll come out losing by a huge margin.

Again, go troll somewhere else.