Sign in with
Sign up | Sign in
Your question

Is Bulldozer Really That Bad?

Tags:
  • CPUs
  • Reviews
  • Bulldozer
  • Product
Last response: in CPUs
Share
a b à CPUs
June 14, 2012 8:35:58 PM

I Was Looking At The FX Price To See If It Went Down More (Im Board) But Read The Reviews For Yourself
http://www.newegg.com/Product/Product.aspx?Item=N82E168...
People Complain Saying Why Does Newegg Let People Review A Product They Don't Have And Everything Says It's Great
And One Person Said It Runs Smoother Then His 2600k No Matter What Benchmarks Say And I Think Benchmarks
Use 256 Bit Floating Points A Lot Which Means It's Comparable To The 2500k But Other Then That
The FX Prosseser Really Seems Smooth To Me And They Are Cheaper Then The 2500k!
I Own A 8120.

More about : bulldozer bad

June 14, 2012 8:46:21 PM

No they are perfectly capable chips, but they are not better than intel alternatives at same price pionts (espeacily for gaming), giving people no reason to buy them.
m
0
l
a c 283 à CPUs
June 14, 2012 8:51:00 PM

You say that because you have one of the two BD chips worth owning IMO. The FX-81xx CPU's are just fine, albeit still not as good as a 2500K (or a high end Phenom II X4, in gaming, for that matter). Where it shines, as I'm sure you know, is in highly threaded applications.

Still, if I was going to buy a FX CPU, a 81xx would be my choice (either that or wait for Piledriver).
m
0
l
Related resources
a c 203 à CPUs
June 14, 2012 8:52:57 PM

It's really not a terrible CPU. It does the job and does it especially well in heavily threaded applications.

If there were no Sandy Bridge, Ivy Bridge or Phenom II X4 / X6 CPUs people would be buying a lot more of them.
m
0
l
a c 203 à CPUs
June 14, 2012 8:54:27 PM

melikepie said:
And One Person Said It Runs Smoother Then His 2600k No Matter What Benchmarks Say
I think that might be down to wishful thinking.
m
0
l
June 14, 2012 9:05:57 PM

WR2 said:

If there were no Sandy Bridge, Ivy Bridge or Phenom II X4 / X6 CPUs people would be buying a lot more of them.


So if there was no competition people would buy them :lol:  well ive doubt of that lol
m
0
l
a c 203 à CPUs
June 14, 2012 9:08:25 PM

It's good to have choices.
m
0
l
a c 154 à CPUs
June 15, 2012 4:10:06 AM

For the people that say it's not bad or is compleatly capable let me ask you this. A Pentium 4 is compleatly capable of physically running any program (not good but it can run any program out there) but do you want one? I don't and I don't want a Bulldozer either.
m
0
l
a b à CPUs
June 15, 2012 4:47:34 AM

WR2 said:
I think that might be down to wishful thinking.

Here was the full review (it's for the 8120). It said he is a verified owner (:

Rating: 5 Stars (or eggs in this case)

beats 2600k

Pros: i have this side by side with a intel 2600k with everything the exact same as far as hardware besides mobo and cpu. this is way smoother than the 2600k. even for vary simple task it is faster. i dont believe any of the benchmarks ive read after buying this cpu. I bought this cpu after the 2600k mind you. its just better. the rig feels smoother and faster. i love my 2600k but i love this 8120 MORE. Great cpu.

Cons: to much power draw.
'Other Thoughts: just buy it or wait for vishera
m
0
l
a c 198 à CPUs
June 15, 2012 5:00:24 AM

melikepie said:
Here was the full review (it's for the 8120). It said he is a verified owner (:

Rating: 5 Stars (or eggs in this case)

beats 2600k

Pros: i have this side by side with a intel 2600k with everything the exact same as far as hardware besides mobo and cpu. this is way smoother than the 2600k. even for vary simple task it is faster. i dont believe any of the benchmarks ive read after buying this cpu. I bought this cpu after the 2600k mind you. its just better. the rig feels smoother and faster. i love my 2600k but i love this 8120 MORE. Great cpu.

Cons: to much power draw.
'Other Thoughts: just buy it or wait for vishera


I don't think he actually has a 2600k
m
0
l
a c 154 à CPUs
June 15, 2012 5:01:16 AM

Lol you call that a review. Lets see some proof that it's so much better than a 2600k. To me the whole "review" sounds more like an AMD fanboys wet dream.
m
0
l
a b à CPUs
June 15, 2012 5:03:29 AM

rds1220 said:
Lol you call that a review. Lets see some proof that it's so much better than a 2600k. To me the whole "review" sounds more like an AMD fanboys wet dream.

I know Bulldozer is weak when it comes to 256 bit floating point numbers but programs that dont use that (most benchmarks do) will run much better then a 2600k.
m
0
l
a c 154 à CPUs
June 15, 2012 5:07:59 AM

A I7 2600k will beat out the Bulldozer in all but a VERY FEW programs. The Bulldozer can only beat out a in highly threaded programs but even then it only beats it out by a small margin. An I7 can still keep up even in heavily threaded programs.
m
0
l
a b à CPUs
June 15, 2012 5:10:09 AM

rds1220 said:
A 2600 I7 will beat out the Bulldozer in all but a VERY FEW programs. Only in highly threaded programs the Bulldozer can beat out an I7 but even at that it only beats out by a small margin. An I7 can still keep up even in heavily threaded programs.

Ok then that's what benchmarks say but for reviewers thats a whole nother story.
m
0
l
a c 198 à CPUs
June 15, 2012 5:12:17 AM

melikepie said:
I know Bulldozer is weak when it comes to 256 bit floating point numbers but programs that dont use that (most benchmarks do) will run much better then a 2600k.


Bulldozer is weak in a lot more than just 256 bit AVX functions. The front end integer resources are just too spread out, forcing threads to be vectorized over multiple cores in order to achieve the same results that a single Sandybridge core can achieve. This puts a lot more stress on the schedulers which drastically reduces performance of applications that have a low number of thread contexts.
m
0
l
June 15, 2012 5:24:27 AM

Speaking from first hand experience. I run my company on KVM-QEMU Debian Testing w/ 4 virtual servers consisting of MySQL Dbase, Forum, Apache2, Remote Desktop hosting, a WIndows Server 2008 as a PDC and I can't believe how fast this thing is. Windows Server 2k8 boots up and shutsdown in under 30 seconds, hell most of the time under 10 seconds.

I think the bad performance is tied to how MS handles threads. Because Linux / Debian is a freaking Monster at the thread management. Windows, regardless if it's XP 32bit, VISTA 64-bit , 7 64-bit, Server 2k3 32 bit or server 2k8 64-bit is easily 10 times faster when run virtually, leaving MS not controlling the threads but the host OS (debian in my case) instead.

I have tried and tried and tried to bog this thing down (8gb of ram which I will never use) and it's never even come CLOSE to it. So the processor itself is a beast. I wouldn't judge the benchmarks based on a Windows Kernel, period.

MS Doesn't write Kernels, it buys them. NT came from IBM (aka OS/2 Warp). The updates to the kernel for VISTA/7 (same kernel mainly)... yep IBM wrote that as MS paid them to do as well.

I have built a Win 7 PRO w/ 16 gb ddr3 ram on the 8150 (same I have above) and it's not a noticeable increase at all. However if I put LInux kernel in charge of handeling the threads, it would fly....

This is first hand, real experience. MS is behind the ball when it comes to multi-core processing / stream processing... BY A LONG SHOT and most of the reviews today are all based off that broken streaming scheme.
m
0
l
June 15, 2012 5:27:08 AM

and for all you gamers out there, if more developers would release games using OpenGL instead of D3D, linux gaming benchmarks would be much much much better than windows as everything in WINE (which lets LInux users use/play windows apps in linux) would kill windows benchmarks.
m
0
l
a c 198 à CPUs
June 15, 2012 6:08:25 AM

antilycus said:
Speaking from first hand experience. I run my company on KVM-QEMU Debian Testing w/ 4 virtual servers consisting of MySQL Dbase, Forum, Apache2, Remote Desktop hosting, a WIndows Server 2008 as a PDC and I can't believe how fast this thing is. Windows Server 2k8 boots up and shutsdown in under 30 seconds, hell most of the time under 10 seconds.


Startup and shutdown times are generally IO constrained, not CPU constrained. My current desktop takes about 15 seconds from power to login, mostly thanks to OPROM intialization. The startup/shutdown routines are also much more complex on Windows than they are on Linux due to all the additional subsystems.

I think the bad performance is tied to how MS handles threads. Because Linux / Debian is a freaking Monster at the thread management. Windows, regardless if it's XP 32bit, VISTA 64-bit , 7 64-bit, Server 2k3 32 bit or server 2k8 64-bit is easily 10 times faster when run virtually, leaving MS not controlling the threads but the host OS (debian in my case) instead. said:
I think the bad performance is tied to how MS handles threads. Because Linux / Debian is a freaking Monster at the thread management. Windows, regardless if it's XP 32bit, VISTA 64-bit , 7 64-bit, Server 2k3 32 bit or server 2k8 64-bit is easily 10 times faster when run virtually, leaving MS not controlling the threads but the host OS (debian in my case) instead.


Virtualization does not improve performance, ever. VTx works by creating a host process for each virtual processor exposed to the guest OS and using VTx instructions to control the state of execution. Inside of that process the guest OS has full context control even if it is virtualization aware. Applications will not run faster in a virtualized Windows environment simply because you have a linux host.

I have tried and tried and tried to bog this thing down (8gb of ram which I will never use) and it's never even come CLOSE to it. So the processor itself is a beast. I wouldn't judge the benchmarks based on a Windows Kernel, period. said:
I have tried and tried and tried to bog this thing down (8gb of ram which I will never use) and it's never even come CLOSE to it. So the processor itself is a beast. I wouldn't judge the benchmarks based on a Windows Kernel, period.


The benchmarks have shown the same results regardless of what platform its being run on. Good parallel support, mediocre linear support.


MS Doesn't write Kernels, it buys them. NT came from IBM (aka OS/2 Warp). The updates to the kernel for VISTA/7 (same kernel mainly)... yep IBM wrote that as MS paid them to do as well.
said:

MS Doesn't write Kernels, it buys them. NT came from IBM (aka OS/2 Warp). The updates to the kernel for VISTA/7 (same kernel mainly)... yep IBM wrote that as MS paid them to do as well.


Whatever you say, fanboyism at work


I have built a Win 7 PRO w/ 16 gb ddr3 ram on the 8150 (same I have above) and it's not a noticeable increase at all. However if I put LInux kernel in charge of handeling the threads, it would fly....
said:

I have built a Win 7 PRO w/ 16 gb ddr3 ram on the 8150 (same I have above) and it's not a noticeable increase at all. However if I put LInux kernel in charge of handeling the threads, it would fly....


So you put Windows 7 on the same processor as your testing box and didn't notice an improvement? I'm seriously confused by this.


This is first hand, real experience. MS is behind the ball when it comes to multi-core processing / stream processing... BY A LONG SHOT and most of the reviews today are all based off that broken streaming scheme.
said:

This is first hand, real experience. MS is behind the ball when it comes to multi-core processing / stream processing... BY A LONG SHOT and most of the reviews today are all based off that broken streaming scheme.


Multiprocessor scaling is very good in Linux, this much is certain. However, on platforms that don't have 64+ hardware threads there's not as much of a difference. The tradeoff is that Microsoft has an enormous userbase with a tremendous range of usecases and Microsoft takes compatibility very seriously. Compatibility in Linux is a huge joke
m
0
l
a b à CPUs
June 15, 2012 6:17:19 AM

antilycus said:
and for all you gamers out there, if more developers would release games using OpenGL instead of D3D, linux gaming benchmarks would be much much much better than windows as everything in WINE (which lets LInux users use/play windows apps in linux) would kill windows benchmarks.

Im learning openGL beacause i dont want to limit people using D3D.
m
0
l
a b à CPUs
June 15, 2012 6:33:32 AM

So back on to the subject of the FX.
You can get one where I live for £50 less than an i5 2500k which I think is very good value. It's nowhere near as good as the i7s and pretty crap next to i5s but that's because it's not in their price ranges so shouldn't have to be.
m
0
l
a b à CPUs
June 15, 2012 7:04:44 AM

Robi_g said:
So back on to the subject of the FX.
You can get one where I live for £50 less than an i5 2500k which I think is very good value. It's nowhere near as good as the i7s and pretty crap next to i5s but that's because it's not in their price ranges so shouldn't have to be.

true :kaola: 
m
0
l
June 15, 2012 10:04:29 AM

rds1220 said:
Lol you call that a review. Lets see some proof that it's so much better than a 2600k. To me the whole "review" sounds more like an AMD fanboys wet dream.


Yep. Sounds about right.
m
0
l
a b à CPUs
June 15, 2012 12:18:43 PM

At this point in time there is no reason at all to buy a bulldozer cpu even if they are cheaper.
m
0
l
June 15, 2012 12:39:44 PM

All Bulldozer is really is a semi-server chip that is decent at running lots of threads but lacking in raw horsepower. When it comes down to it the decision on what you buy is based on previous prejudices.

@Pinhedd: Pretty much ALL consumer hardware is made to run in windows, who do you think they are selling to? Linux is extremely compatible when you decide to recognize the fact that compatibility in most Linux OSs is built in by unpaid programmers who do it in their spare time(usually).

m
0
l
a c 487 à CPUs
June 15, 2012 1:41:41 PM

Well... in the big picture 88% of the 387 reviews for the FX-8150 were at least 4 stars. Pretty good actually.

But... out of the 2,073 reviews for the Core i5-2500k, 98% of them were for at least 4 stars. And one bad review (1 star) was directed at an Asus P67 motherboard that came bundled with the Core i5-2500k which apparently died.

http://www.newegg.com/Product/Product.aspx?Item=N82E168...
m
0
l
a b à CPUs
June 15, 2012 1:59:00 PM

antilycus said:

I think the bad performance is tied to how MS handles threads. Because Linux / Debian is a freaking Monster at the thread management. Windows, regardless if it's XP 32bit, VISTA 64-bit , 7 64-bit, Server 2k3 32 bit or server 2k8 64-bit is easily 10 times faster when run virtually, leaving MS not controlling the threads but the host OS (debian in my case) instead.


Wrong. When you virtualize an OS, you virtualize EVERYTHING, including the thread management. Virtualizing windows doesn't magically change the way the scheduler work.

Quote:

MS Doesn't write Kernels, it buys them. NT came from IBM (aka OS/2 Warp). The updates to the kernel for VISTA/7 (same kernel mainly)... yep IBM wrote that as MS paid them to do as well.


Wrong. OS/2 shares more simmilarity with the old Win9x kernel; the NT based kernel was written by MS in parallel, coincidentally by the guys who wrote the VAX operating system. Thats one reason why NT, for the most part, is stable.

Secondly, the OS/2 Kernel [prior to the MS/IBM split] was mostly written by MS, not IBM.

Quote:

I have built a Win 7 PRO w/ 16 gb ddr3 ram on the 8150 (same I have above) and it's not a noticeable increase at all. However if I put LInux kernel in charge of handeling the threads, it would fly....


Which linux scheduler are you using? Theres 5 or 6 that I know of, and you can easily use one to optimize certain benchmarks [IE: Theres a low priority one, and equal oppertunity one, a priority based one, etc]. All schedulers are not created equal; you have to pick and choose what type of tasks you want to be best at.

The way the Windows scheduler works is probably the "best" implementation possible if you have to go with ONE scheduler: The highest non-I/O blocked thread ALWAYS executes. Key windows API's get a priority boost, as do forground applications [as of Vista].

Quote:

This is first hand, real experience. MS is behind the ball when it comes to multi-core processing / stream processing... BY A LONG SHOT and most of the reviews today are all based off that broken streaming scheme.


Its not MS's fault; people need to understand that serial processes can not, and will not, ever scale.
m
0
l
a c 198 à CPUs
June 15, 2012 7:09:13 PM

AhabTheWhaler said:
All Bulldozer is really is a semi-server chip that is decent at running lots of threads but lacking in raw horsepower. When it comes down to it the decision on what you buy is based on previous prejudices.

@Pinhedd: Pretty much ALL consumer hardware is made to run in windows, who do you think they are selling to? Linux is extremely compatible when you decide to recognize the fact that compatibility in most Linux OSs is built in by unpaid programmers who do it in their spare time(usually).


I wasn't referring exclusively to hardware compatibility, I'm well aware that much hardware is rendered incompatible by manufacturers keeping their specifications closed which forces developers to reverse engineer the Windows drivers.

Source and binary compatibility can be equally as awful, especially when kernel driver interfaces keep changing. It's gotten better though.
m
0
l
a b à CPUs
June 15, 2012 8:09:05 PM

jaguarskx said:
Well... in the big picture 88% of the 387 reviews for the FX-8150 were at least 4 stars. Pretty good actually.

But... out of the 2,073 reviews for the Core i5-2500k, 98% of them were for at least 4 stars. And one bad review (1 star) was directed at an Asus P67 motherboard that came bundled with the Core i5-2500k which apparently died.

http://www.newegg.com/Product/Product.aspx?Item=N82E168...

Well I Get The Point That People Liked The 2500k But My Point Is That People Not Just Liked The FX Processors But They Said That The Benchmarks Seemed Wrong And They Liked The Chip!
m
0
l
a b à CPUs
June 15, 2012 8:40:01 PM

melikepie said:
Well I Get The Point That People Liked The 2500k But My Point Is That People Not Just Liked The FX Processors But They Said That The Benchmarks Seemed Wrong And They Liked The Chip!

http://en.wikipedia.org/wiki/Cognitive_dissonance
m
0
l
August 27, 2012 7:53:11 PM

Pinhedd said:
Virtualization does not improve performance, ever. VTx works by creating a host process for each virtual processor exposed to the guest OS and using VTx instructions to control the state of execution. Inside of that process the guest OS has full context control even if it is virtualization aware. Applications will not run faster in a virtualized Windows environment simply because you have a linux host.


I can think of one scenario. Let's say you have a client VM and server VM living on the same host. The network I/O between both of those VMs should be far better than compared to two physical servers.
m
0
l
a c 154 à CPUs
August 27, 2012 10:54:14 PM



AKA Fanboyism. A person will claim the Bulldozer is the greatest thing since the invenion of the PC. They know it's slow and junky but they'll do anything to defend their buy.
m
0
l
!