Compare 1090T and FX-8150

[eRazorzEDGE]

I was curious of how my old processor would compete against the new one, so I tested them both and well, here are the results.

Here's a lil info on the system. The system is mounted on a custom built aluminum test bench. The CPU cooling is done by a single Feser 120.3 radiator with three Scythe 120mm x 25mm fans running at 1500rpm. The pump is a Laing D5 Vario at max speed. The water block is a Swiftech GTZ with custom 0.25" aluminum back plate and tension springs.

I wanted to see what the difference in performance would be between these two chips as I would run them 24/7. So I OC'ed them both to there maximum semi-stable frequencies and off we went. Basically if they passed Cinebench and 3Dmark CPU tests, I considered it stable. The 1090T is at 4.3GHz Core @ 1.55v, 2.6GHz CPU-NB @ 1.25v and 2.6GHz HT Link @ 1.20v. The FX-8150 is at 4.8GHz Core @ 1.45v, 2.6GHz CPU-NB @ 1.25v and 2.6GHz HT Link @ 1.20v. Both CPU's were run with the same RAM, 2x4GB DDR3-1600MHz running at 1600MHz. The video cards are 3x XFX 6970's running at stock frequencies and voltages (880MHz core, 1375MHz RAM). I forgot to add a GPU-Z shot in the screenies, so forgive me. Also, I used a different CPU-Z from the 1090T to the FX-8150 because the 1.58 version didn't show the correct information of the FX, like the 1.58.7 version does.

The test system:






SuperPi - 1090T (first), FX-8150 (second)



wPrime - 1090T (first), FX-8150 (second)



Cinebench - 1090T (first), FX-8150 (second)



Unigine Heaven (1920x1200, DX11, Shaders = High, Tess. = Normal, Anisotropy = 16x, AA = Off) - 1090T (first), FX-8150 (second)



3Dmark 06 - 1090T (first), FX-8150 (second)



3Dmark 11 - 1090T (first), FX-8150 (second)



------------------------------------------------------------------------------------------------------------------------
-EDIT-
BELOW ARE MORE GAME BENCHES AS OF 10-23-11

I did a lil more testing with the FX-8150 at 4.8GHz CPU, 2.6GHz NB and HT Link, and 1600MHz RAM. After that I put the 1090T back in and for some stupid reason I couldn't boot at 2.6GHz on the NB and HT Link, so I upped the frequencies to 4.35GHz CPU (was 4.3GHz), 3.0GHz NB and HT Link (was 2.6GHz), and 1600MHz RAM. I used the same entire setup as before, with 3x 6970's in Tri-Fire running at stock frequencies of 880MHz core and 1375MHz memory, ASUS Crosshair V mobo running the beta 9911 BIOS, water cooled CPU with a single 360 RAD and fans at 1500RPM.

The test games I chose are ones that I play or have played. Two of which have in game benchmarks, the other two I had to run a FRAPS 60 second capture for min-max-avg frames per second. The games are Battlefield: Bad Company 2, Crysis, Dirt 3, and Left 4 Dead 2. I did two runs of each game, one at 1920x1200 and the other at 5760x1200. All games were run at their maximum in game graphic settings with either 8xAA, 4xAA, or no AA (crysis only).

I had originally intended to do some encoding with handbrake, but for the life of me I couldn't get the FX-8150 to stay stable for a whole minute under load, so only game benches here.

Battlefield: Bad company 2 - All settings on High, 4xAA, 16xAF, HBAO: On - Captured during a multi-player match on the map Laguna Presa.

X6 1090T - 1920x1200:
Frames, Time (ms), Min, Max, Avg
6764, 60000, 74, 201, 112.733

X6 1090T - 5760x1200:
Frames, Time (ms), Min, Max, Avg
3049, 60000, 37, 84, 50.817

FX-8150 - 1920x1200:
Frames, Time (ms), Min, Max, Avg
5424, 60000, 70, 114, 90.400

FX-8150 - 5760x1200:
Frames, Time (ms), Min, Max, Avg
2942, 60000, 37, 72, 49.033

Crysis - All settings on Very High, 0xAA - Used the Crysis Benchmark Tool v1.05

X6 1090T - 1920x1200:


X6 1090T - 5760x1200:


FX-8150 - 1920x1200:


FX-8150 - 5760x1200:


Dirt 3 - All settings on Ultra, 4xAA - Used the in game benchmark tool set to the Aspen map.

X6 1090T - 1920x1200:


X6 1090T - 5760x1200:


FX-8150 - 1920x1200:


FX-8150 - 5760x1200:


Left 4 Dead 2 - All settings maxed out, 8xAA - Played single player campaign on the map "The Sacrfice: 1. Docks", started capture from when I was first able to move the character.

X6 1090T - 1920x1200:
Frames, Time (ms), Min, Max, Avg
12215, 60000, 138, 285, 203.583

X6 1090T - 5760x1200:
Frames, Time (ms), Min, Max, Avg
7530, 60000, 99, 160, 125.500

FX-8150 - 1920x1200:
Frames, Time (ms), Min, Max, Avg
14312, 60000, 159, 293, 238.533

FX-8150 - 5760x1200:
Frames, Time (ms), Min, Max, Avg
7171, 60000, 91, 143, 119.517

[BeepBeep2]

This is pretty awful.
Are you able to bench at 5 Ghz?

...add two "cores", add 500 Mhz, 32nm process, get lower or much less efficient results.

[eRazorzEDGE]

Not very effectively. The chip gets really hot cuz I need 1.55v to run it at 5GHz and do multiple 100% load benches. Plus it's just randomly crashing and rebooting at the strangest times, like looking at web pages, browsing files on my drives, and just idling while I sit here and think of stuff to type. It's quite aggravating.

Yeah, it's pretty sad that it can't even out gun it's own previous generation chip. But there is a bright side, if you make the GPU's do all the calculations, the CPU looks really good at 4.8GHz.

*edit*
Oh, and it's gotta be draining power off the grid, at idle with the 1090T, I would see about a 1-2° delta between air and water temps. With the FX-8150, I'm now seeing a 6° delta.

[Manicdan]

i dont consider 1.55v on thuban 24/7 for most people. most dont push it past 1.5v because air cooling just cannot keep it stable around there (unless you have a fan that makes you deaf)

also the crashing in idle ive seen before, and its because my idle voltage was too low. i would turn that up a few notches or just turn off CnQ and see if that fixes it.

can you test power consumption with the 2 different chips? it would be great to see identical load wattage and see how different the temps are, since that can give a very clear indication if the cpus are set with different calibrated sensors.

[BeepBeep2]

i dont consider 1.55v on thuban 24/7 for most people. most dont push it past 1.5v because air cooling just cannot keep it stable around there (unless you have a fan that makes you deaf)

also the crashing in idle ive seen before, and its because my idle voltage was too low. i would turn that up a few notches or just turn off CnQ and see if that fixes it.

can you test power consumption with the 2 different chips? it would be great to see identical load wattage and see how different the temps are, since that can give a very clear indication if the cpus are set with different calibrated sensors.

You know what the bad thing about FX is?
You should get a 30% performance/watt increase with the move to 32nm vs 45.

If you produced an FX-6100 (33% less cores than FX-8150) on 45nm it would have the same power consumption as X6 1100T and lose to it in EVERY benchmark AND real-world task. Even overclocked, it would lose in EVERY benchmark. ...that is, excluding benchmarks that take advantage of the new instructions which are little to none currently.

[mav2000]

Thanks for this comparison....this is really getting worse and worse. I thought the difference between the old and new was a little, but IPC seems to be really really bad on this new chip. As a request, could you try the same tests with 4c/4cu...?

[Swanie]

Can you do some gaming comparisons please, as i am looking getting the BD for gaming.

[PatRaceTin]

Thank i wait for FX hit retail in Thailand

[zanzabar]

is the asus bios still nerffing the chip when compared to the asrock?

[FlanK3r]

Il show sometihng more diferent with OC Thuban vs OC FX...Maybe, u will be suprised....give me some weeks (2-3), this week Im busy and I have no much time for HW.

[Swanie]

@Flanker_ are you happy with your chip?

[Zeus]

Thanks for this nice compare, this gives us a good insight in how superior FX really is.

One thing though, did you use the same mobo w/ same bios for the tests? I assume it was the Asus Crosshair V? What happens if you put the Thuban in a Crosshair IV w/ Thuban optimised bios, the difference could be even more.

Oh that 3d mark06 result really shocked me btw!

[eRazorzEDGE]

i dont consider 1.55v on thuban 24/7 for most people. most dont push it past 1.5v because air cooling just cannot keep it stable around there (unless you have a fan that makes you deaf)

also the crashing in idle ive seen before, and its because my idle voltage was too low. i would turn that up a few notches or just turn off CnQ and see if that fixes it.

can you test power consumption with the 2 different chips? it would be great to see identical load wattage and see how different the temps are, since that can give a very clear indication if the cpus are set with different calibrated sensors.

You'll notice after I said "24/7", I clarified by saying "... maximum semi-stable frequencies... if they passed Cinebench and 3Dmark CPU tests, I considered it stable." For me, if the rig can stay stable w/o crashing for a 10-hour game session, I call it 24/7 stable.

I think I forgot to mention that I had all power saving options disabled in the BIOS. I also had the HPC (High Performance Computing) option turned off, but I really have no idea what the hell that does. Anybody have a clue?

I'd love to test power consumption, but the last Kill-o-watt I had melted to the receptacle and I haven't got one since.

Thanks for this comparison....this is really getting worse and worse. I thought the difference between the old and new was a little, but IPC seems to be really really bad on this new chip. As a request, could you try the same tests with 4c/4cu...?

I was planning on testing that out, too. Right now I'm working on the random crashing, but I'll get to it soon.

Can you do some gaming comparisons please, as i am looking getting the BD for gaming.

Give me some games that I can easily bench. I usually play Bad Company 2, but I don't think there's any pre-configured benchmark runs. I suppose I could capture the frames with Fraps and try to make the same run twice.

is the asus bios still nerffing the chip when compared to the asrock?

Oh, I'm also running the 9911 BIOS, since I didn't mention it before. But as far as this board being better/worse than another, I couldn't tell ya. I've seen the same benches as everyone else, but I find it hard to accept that the flagship ASUS mobo is going to suck by that much, compared to other boards.

[eRazorzEDGE]

Thanks for this nice compare, this gives us a good insight in how superior FX really is.

One thing though, did you use the same mobo w/ same bios for the tests? I assume it was the Asus Crosshair V? What happens if you put the Thuban in a Crosshair IV w/ Thuban optimised bios, the difference could be even more.

Oh that 3d mark06 result really shocked me btw!

Same mobo, same 9911 Beta BIOS. Oops, I guess I forgot to mention the board I was running, heh. It is the ASUS Crosshair V. The difference between the CHV and CHIV is negligible. The 3dmark scores are near identical to those I've posted on the ORB with the CHIV. Same with the Cinebench.

[FlanK3r]

@Flanker_ are you happy with your chip?

yes, after OC is it really better than OC Thuban in practice. Interesting is some as megatasking, FX is more quickly better than Thuban chip (perhaps because FX has more better IMC). Il test for comparsion max stable OC Thuban vs FX in practice compression, decompression, videoencoding, practice 3DsMax rendering, Cinebenchs, practice Photoshop, CAD apliactions etc. Think, it will be interesting

[Leeghoofd]

Same mobo, same 9911 Beta BIOS. Oops, I guess I forgot to mention the board I was running, heh. It is the ASUS Crosshair V. The difference between the CHV and CHIV is negligible. The 3dmark scores are near identical to those I've posted on the ORB with the CHIV. Same with the Cinebench.

Wouldn't it have been better even to test the 1090T with an older bios, not for BD optimised one ?

[BeepBeep2]

Thanks for this nice compare, this gives us a good insight in how superior FX really is.

One thing though, did you use the same mobo w/ same bios for the tests? I assume it was the Asus Crosshair V? What happens if you put the Thuban in a Crosshair IV w/ Thuban optimised bios, the difference could be even more.

Oh that 3d mark06 result really shocked me btw!

On 0051 this board loses about 15 seconds in 32M with Thuban @ 4.5, and BIOSes after that were more BD optimized... so lets just call that a 3.5% hit.

The difference in score is pretty rediculous. I assume Thuban is losing score in wPrime + Cinebench as well, if it's less or more than 3.5% I don't know but 3.5% is the equivilent of 150 Mhz loss in efficiency at 4.5.

[Swanie]

@ eRazorzEDGE: Thanks for your testing and feedback, highly appreciated. I also play BFBC 2 , so any average FRAPS feedback will do,then also how does the gameplay feel in comparison to the 2 chips: any big drops in FPS that impacts on gameplay as i see you playing on eyefinity also.
@FlanK3r: Thank you also. It is truly nice to see day to usage reports from guys actually who owns these chips.

I took the plunge placed my order for mobo and chip. sold my I7 as per sig.

Looking forward to more feedback and inputs from all

[RCG_Bex]

Word on the street is the Asus BIOS for BD isn't helping. Some tests over on the ASRock boards show BD improve.

[Zeus]

On 0051 this board loses about 15 seconds in 32M with Thuban @ 4.5, and BIOSes after that were more BD optimized... so lets just call that a 3.5% hit.

The difference in score is pretty rediculous. I assume Thuban is losing score in wPrime + Cinebench as well, if it's less or more than 3.5% I don't know but 3.5% is the equivilent of 150 Mhz loss in efficiency at 4.5.

The good news is that we can keep our money in our pockets. I think i'm gonna invest in a nice SSD instead. It would have been nice if i could just have dropped a BD cpu in this Crosshair IV F but i have a gut feeling it ain't gonna happen.

Word on the street is the Asus BIOS for BD isn't helping. Some tests over on the ASRock boards show BD improve.

Can you believe AMD choose this CVF as a reference board if this is true?

[Manicdan]

You know what the bad thing about FX is?
You should get a 30% performance/watt increase with the move to 32nm vs 45.

thuban is the end of 45nm and this is sorta the beginning of 32nm
i would compare efficiency of the PII 940 to FX 8150 and see how much that changed.

[Yann_H]

Thanks for your Bench!

I think i'll keep my 1090T... This FX 8150 doesn't give enough for the money...

Do you know if we can run it (FX 8150) on crosshair IV formula? I have seen on asus website a bios update to support AM3+ cpu but I have also read that the socket is different including "bigger" holes for pins...

[demonkevy666]

Thuban 64Kbytes x6 vs 64Kbytes x4 Bulldozer (128Kbytes less) instruction cache.

[Mechromancer]

is the asus bios still nerffing the chip when compared to the asrock?

I'm really interested in where this BIOS fiasco I'm reading about on multiple sites leads us. Are there any more tests with the Asrock?

[Swanie]

Word on the street is the Asus BIOS for BD isn't helping. Some tests over on the ASRock boards show BD improve.

Ye i read that the ASRock performs much better with the BD, pity here in South Africa we cant get this mobo so i am stuck with MSI, Asus or Giga and going for the UD7