Can AMD "rescue" the Bulldozer?

[LOL_Wut_Axel]

Intel has more than two CPUs (2500K/2600K) :whiste:

Bellow the $200,00 mark, AMDs Bulldozer is faster if Overclocked. Well unfortunately Intel have chosen not to allow all its CPUs to be Overclockable. :thumbsdown:

A Hex core AMD FX6100 priced at $149,99 overclocked to 4,7GHz will be faster and cheaper than Intel's Core i5 2310 priced at $184,99 that can only OC up to 3,2GHz.

The price difference can go to a faster graphics card, from HD6850 to HD6870 or from GTX560 to GTX560 ti, or for faster/more memory or better Motherboard.

So yes, 2500K/2600K are nice CPU's but not everyone can afford them

And your point is? A Phenom II X6 1055T at 4-4.2GHz is faster than an FX-6100 at 4.5-4.7GHz, so your point is moot. Also, talking about getting faster graphics cards and then using AMD CPUs is pretty much an oxymoron. If you want a good gaming CPU on the cheap you get a Core i3-2120 instead. If it's for multi-threaded applications you buy a 1055T instead. Once again, against AMD's own previous efforts, Zambezi sucks. Against Intel's latest efforts, it also sucks.

 

[Phynaz]

Intriguing, but not likely.

Jeez, the nuke from orbit post. Didn't want the debate to go on just a little, did you? :biggrin:

 

[sequoia464]

You still haven't provided any proof yet that 4.7Ghz is common on a 6100

I'm just a casual user here, had an AM3+ board when BD was released so I figured it was worth checking out a BD. I have a 6100 ($160 when I bought it) set to 4200MHz - 1.375v ... with turbo enabled to 4700 - 1.4125v.

Mine runs fine for me, I don't game however, My usage involves some fairly large Tiff's, XML files and smaller excel files.

There is a fairly large BD owners thread at Overclockers.net, it is not at all unusual to see people running their 6100's at 4700 or above, I don't know if this is representative of what they are capable of en mass or not - mine is on a work machine so I am conservative with my settings.

EDIT: In hindsight however I would have gone with a 2500K

 

[Phynaz]

Thanks, that's what I was looking for.

So we can now say that a reasonable number of 6100's will turbo to 4.7Ghz (you didn't mention what they were running for base clocks).

Now we just need to bench one against a non-K i5.

 

[AtenRa]

You still haven't provided any proof yet that 4.7Ghz is common on a 6100, or that it would be faster than the i5.

Every BD will run at 4.7GHz at 1.4/1.42V stable

for the unbelievers

 

[Phynaz]

You're a little behind the curve there, we already established that fact.

 

[AtenRa]

Anyone with a Core i5 and a HD6950 1 or 2GB to test a few games at 1080p ??? i can run the FX 8150 with 2 or 3 modules, 4 or 6 threads.

 

[sequoia464]

Thanks, that's what I was looking for.

So we can now say that a reasonable number of 6100's will turbo to 4.7Ghz (you didn't mention what they were running for base clocks).

Now we just need to bench one against a non-K i5.

I think most of the people over there have disabled Turbo. I only found twelve scores validated by CPU-Z - on the first page of the thread - ranged from 3925 to 5105 - averaged 4439MHz. There were others but they were taken when the CPU's were idling.

Twelve scores is not a large sample by any means though - there are more in the thread that people haven't validated, but there are 314 pages in the thread - not something I want to look through.

 

[sequoia464]

Every BD will run at 4.7GHz at 1.4/1.42V stable

for the unbelievers

Do you disable Turbo?

 

[AtenRa]

Do you disable Turbo?

yes

 

[skipsneeky2]

Well, even if you're using it for working (in other words, in multi-threaded applications) it's no better than the cheaper and much more power conservative i5-2500K.

When it comes down to scenarios that use all cores/threads, 8 Zambezi cores=4 Sandy Bridge cores.

Even though it performs the same in multi-threaded, it consumes 2x more power than the i5-2500K.

What you end up with is that Sandy Bridge delivers 2x better performance/watt.

Now please explain to me in what scenario Zambezi makes sense.

LOL the processor doesn't make sense period but maybe to the desperate ocer going from like a dual core athlon who already has a 990fx motherboard by chance but yeah nice damage control with those numbers i tell you they went all out with nothing to brag about

 

[Chiropteran]

Oh, OK, I was mistaken. I thought you were defending the fact that BD sucks. You're just saying that it's better than Llano, not that it is any good in general.

Thanks for the clarification.

Absolutely. I wouldn't recommend bulldozer to anyone except under very specific conditions, such as someone who already has a compatible motherboard, doesn't need single threaded performance, uses apps which actually run well on bulldozer, and doesn't care about power usage.

 

[AtenRa]

Ok,
I have a few benchmark runs i did in the past for another review im currently working on with Intel Core i7 2600K with HT Off (4 threads), Turbo Off at default frequency of 3,4GHz, memory at 1600MHz.
I know Core i7 has more L3 cache than i5 but I will use those runs to simulate the Core i5 at 3.4GHz, we will get an idea of the performance.

I have just run the same benchmarks with FX8150 but only with 3 Modules 6 Threads at 4,77GHz, Turbo Off, memory at 1666MHz.

Let’s have it,

Χ264 Benchmark (higher is better)
AMD FX 6100 First Pass = 143,58 fps - Second Pass = 36,60 fps
Intel Core i7 2600K@ First Pass = 148,13 fps - Second Pass = 29,16 fps

7zip (Higher is Better)
AMD FX 6100 Compression = 18390 MIPS - Decompression = 21075MIPS
Intel Core i7 2600K Compression = 14925 MIPS - Decompression = 14394MIPS

POV-Ray v3.7 RC3 (Balcony)
AMD FX 6100 = 14298pps/55sec
Intel Core i7 2600K = 13107pps/1m

Cinebench 11.5 64-bit (multithread)
AMD FX 6100 = 5,78
Intel Core i7 2600K = 5,35

AMD FX-6100 Overclocked at 4,77GHz is faster than Intel Core i5 at 3,4GHz.
Because I don’t have the Core i7 now, I can’t bench Games. I will try to use my cousin Core i5 2500K PC in a few days and run more benchmarks.

Edit: Those runs are without the latest Microsoft BD putch

 

[Phynaz]

I wonder why your 2600K numbers are so much lower than Anands. Hmmmmm....

 

[Arzachel]

I wonder why your 2600K numbers are so much lower than Anands. Hmmmmm....

Both HT and Turbo is off.

 

[AtenRa]

http://www.anandtech.com/bench/Product/288?vs=434

Anands Core i5 2500K Cinebench 11.5 = 5,42
Since my run (5,35) was without turbo i find my numbers to be very close

Anands x264 HD Bench v3.03 2nd Pass = 28,7
My run at 26,16 suggest the larger L3 cache of the 2600K make it perform better in this test.
Just to point out i have used version v4.0.

For POV-Ray i have used the Balcony project, since i dont know witch project Anand is using we cannot compere them directly.

And for 7zip my score is higher than Anands.

You have accepted my Cinebench picture before as a proof and now you dont accept my runs ??
If anyone has a Core i5 please run the same benchmarks and post your numbers.

 

[BLaber]

AMD Reality Check at FX GamExperience , Source @ http://legitreviews.com/article/1838/1/

The Pepsi Challenge has been an ongoing marketing promotion run by PepsiCo since 1975. It all started when a Pepsi employee setup a table with two blank cups: one containing Pepsi and one with Coca-Cola. Shoppers were then encouraged to taste both colas, and then select which drink they prefer. Then the representative then revealed what brands were in each cup. At the end of the day the public was doing a blind taste test and the company was getting a consensus of what brand was preferred by more consumers. Over the years this test has been copied by thousands and today AMD is putting their own twist to the challenge down in Dallas, Texas for the FX GamExperience

The first test that the gamers were issued was down on a pair of $500 gaming machines. System A was powered by an Intel Core i3-2105 ' Sandy Bridge' processor with Intel HD Graphics 3000 and an ASRock H61 motherboard. System B was powered by an AMD A8-3850 'Llano' APU with Radeon HD 6550D on an ASRock A55 motherboard. All of the rest of the components were identical. The obvious goal of this test was to look at similarly priced systems and to see if there was a noticeable difference in gaming performance due to the integrated graphics.

The AMD Reality Check Results:

System A (Intel Core i3-2105): 5 Votes
System B (AMD A8-3850): 136 Votes
No Difference: 2 Votes

The next challenge given to gamers was two high-end systems that were both running AMD Radeon HD 7970 'Tahati' DirectX 11 graphics cards running an Eyefinity display setup. The Intel system was powered by an Intel Core-i7 2700K 'Sandy Bridge' processor with an ASRock P67 Fatal1ty motherboard and 8GB of AMD DDR3 performance memory. The AMD system was powered by the FX-8150 'Bulldozer' processor an ASRock 990FX Fatal1ty and the same 8GB of AMD DDR3 performance memory. The key to this demo was focused on processor performance and not graphics performance. The Intel Core i7-2700K retails for $369.99 and the AMD FX-8150 retails for $269.99, so the question here was if gamers could tell a difference between the systems.

The AMD Reality Check Results:

System C (Intel Core i7-2700K): 40 Votes
System D (AMD FX-8150): 73 Votes
No Difference: 28 Votes

 

[gramboh]

The Pepsi Challenge has been an ongoing marketing promotion run by PepsiCo since 1975. It all started when a Pepsi employee setup a table with two blank cups: one containing Pepsi and one with Coca-Cola. Shoppers were then encouraged to taste both colas, and then select which drink they prefer. Then the representative then revealed what brands were in each cup. At the end of the day the public was doing a blind taste test and the company was getting a consensus of what brand was preferred by more consumers. Over the years this test has been copied by thousands and today AMD is putting their own twist to the challenge down in Dallas, Texas for the FX GamExperience

The first test that the gamers were issued was down on a pair of $500 gaming machines. System A was powered by an Intel Core i3-2105 ' Sandy Bridge' processor with Intel HD Graphics 3000 and an ASRock H61 motherboard. System B was powered by an AMD A8-3850 'Llano' APU with Radeon HD 6550D on an ASRock A55 motherboard. All of the rest of the components were identical. The obvious goal of this test was to look at similarly priced systems and to see if there was a noticeable difference in gaming performance due to the integrated graphics.

The AMD Reality Check Results:

System A (Intel Core i3-2105): 5 Votes
System B (AMD A8-3850): 136 Votes
No Difference: 2 Votes

The next challenge given to gamers was two high-end systems that were both running AMD Radeon HD 7970 'Tahati' DirectX 11 graphics cards running an Eyefinity display setup. The Intel system was powered by an Intel Core-i7 2700K 'Sandy Bridge' processor with an ASRock P67 Fatal1ty motherboard and 8GB of AMD DDR3 performance memory. The AMD system was powered by the FX-8150 'Bulldozer' processor an ASRock 990FX Fatal1ty and the same 8GB of AMD DDR3 performance memory. The key to this demo was focused on processor performance and not graphics performance. The Intel Core i7-2700K retails for $369.99 and the AMD FX-8150 retails for $269.99, so the question here was if gamers could tell a difference between the systems.

The AMD Reality Check Results:

System C (Intel Core i7-2700K): 40 Votes
System D (AMD FX-8150): 73 Votes
No Difference: 28 Votes

The first result makes sense ($500 system) but the high end one should be a wash, both experiences should be nearly identical, doesn't make sense for the 8150 to double the votes at all unless there was a config issue.

 

[Chiropteran]

The first result makes sense ($500 system) but the high end one should be a wash, both experiences should be nearly identical, doesn't make sense for the 8150 to double the votes at all unless there was a config issue.

I've always noticed certain computers, such as Dell Intel based Optiplex at work, often "feel" slower than they actually are. Maybe it's a flaw in the BIOS, or some cheap hardware corner they cut, perhaps the cheap slow hard drives. The computers just feel less responsive then they should be, given the CPU. In actual benchmarks they would perform more or less as they should, it was only some slight GUI lag, but it was very noticeable to me.

I suspect that this is the case in many computers, in that apparent speed is affected by a lot of non-CPU factors, and a CPU which benchmarks better may still "feel" slower in manual use. I recall posts in this forum a few years back about old Athlon or Phenom systems "feeling" faster than Intel CPUs, even when the Intel benchmarks better and should be faster.

Perhaps this is something similar, where the FX-8150's features or the motherboards they paired it with just had some responsiveness advantage over the Intel system.

 

[Olikan]

The next challenge given to gamers was two high-end systems that were both running AMD Radeon HD 7970 'Tahati' DirectX 11 graphics cards running an Eyefinity display setup . The Intel system was powered by an Intel Core-i7 2700K 'Sandy Bridge' processor with an ASRock P67 Fatal1ty motherboard and 8GB of AMD DDR3 performance memory. The AMD system was powered by the FX-8150 'Bulldozer' processor an ASRock 990FX Fatal1ty and the same 8GB of AMD DDR3 performance memory.

since when cpu matters in these resulotion?

meh, i bet even starcraft 2 will be gpu bottleneck.

 

[Chiropteran]

since when cpu matters in these resulotion?

meh, i bet even starcraft 2 will be gpu bottleneck.

I do find it funny how the responses change.

A few pages back "Don't know who would be smoking enough crack to purchase a 8 core bulldozer to game."

 

[Idontcare]

The Pepsi Challenge has been an ongoing marketing promotion run by PepsiCo since 1975. It all started when a Pepsi employee setup a table with two blank cups: one containing Pepsi and one with Coca-Cola. Shoppers were then encouraged to taste both colas, and then select which drink they prefer. Then the representative then revealed what brands were in each cup. At the end of the day the public was doing a blind taste test and the company was getting a consensus of what brand was preferred by more consumers.

My anecdotal story does not invalidate the main thrust of your argument, please don't take it that way. I actually got to partake in one of those Pepsi Challenges in a grocery store (not the original though I'm not that old ) and was crazy rigged.

The tester person was merely alternating between the two cups as to which was Coke and which was Pepsi, so everyone in line knew exactly which was which when their turn came up because all they had to do was notice the results of the person in front of them. And the tester of course told/showed you the answers when you were done with your taste test.

But, and this was critical, if you said the pepsi cup was better then you got a free can of pepsi. If you said coke was better then you got nothing. I liked coke better than pepsi, but I wasn't going to turn down a free pepsi, so I voted Pepsi as did about 90% of every one else in the line.

Unless the tests you referenced were done by a 3rd party organizer then I would not trust the results, personally speaking. Who knows what body language was being used by the AMD coordinators, unconsciously.

That said, it is baffling that the results weren't even more in favor of the Llano setup, how could anyone game in a side-by-side with in Intel GPU and feel like it gave the superior gaming experience? That's hanging-chad goofy voting right there.

 

[sequoia464]

That's hanging-chad goofy voting right there.

Can we get a recount??

 

[exar333]

My anecdotal story does not invalidate the main thrust of your argument, please don't take it that way. I actually got to partake in one of those Pepsi Challenges in a grocery store (not the original though I'm not that old ) and was crazy rigged.

The tester person was merely alternating between the two cups as to which was Coke and which was Pepsi, so everyone in line knew exactly which was which when their turn came up because all they had to do was notice the results of the person in front of them. And the tester of course told/showed you the answers when you were done with your taste test.

But, and this was critical, if you said the pepsi cup was better then you got a free can of pepsi. If you said coke was better then you got nothing. I liked coke better than pepsi, but I wasn't going to turn down a free pepsi, so I voted Pepsi as did about 90% of every one else in the line.

Unless the tests you referenced were done by a 3rd party organizer then I would not trust the results, personally speaking. Who knows what body language was being used by the AMD coordinators, unconsciously.

That said, it is baffling that the results weren't even more in favor of the Llano setup, how could anyone game in a side-by-side with in Intel GPU and feel like it gave the superior gaming experience? That's hanging-chad goofy voting right there.

Might be reading comprehension on my part, but didn't Llano destroy the HD3000 in votes per the quoted text? Maybe my sarcasm meter is also broken...

The AMD Reality Check Results:

System A (Intel Core i3-2105): 5 Votes
System B (AMD A8-3850): 136 Votes
No Difference: 2 Votes

 

[Chiropteran]

Might be reading comprehension on my part, but didn't Llano destroy the HD3000 in votes per the quoted text? Maybe my sarcasm meter is also broken...

The AMD Reality Check Results:

System A (Intel Core i3-2105): 5 Votes
System B (AMD A8-3850): 136 Votes
No Difference: 2 Votes

I think what Idontcare was saying was that the 5 votes for system A were goofy. My guess is it was some non-gamers who felt compelled to vote and just kinda picked one like "oh I dunno this first one is faster, yep" when they didn't even know what they were looking for.