BitFenix Fury Gold 750 W

[crmaris]

Bitfenix decided to enter the PSU market, and today, we have the chance to look at the flagship of their Fury series. It has a capacity of 750 W, is 80 Plus Gold certified, and features individually sleeved cables of high quality and stunning looks.

Show full review

 

[GreiverBlade]

stunning look, nice sleeving ... horrible price (stiff is really the word for it ... )
the 8.3 rating is well deserved for a 1st attempt from them

i think if the price of the 550/650 would be under 120$ i would swap my Thermaltake Smart SE 530w but that Tt is less than 65$ where i live and the main cables and modular cables are all black flat type (which is kinda a nice addition at that price point)

anyway nice review

 

[crmaris]

as they told me these cables are really expensive (to make obviously) so the price went up for at least 30-40$ only because of them.

 

[GreiverBlade]

as they told me these cables are really expensive so the price went up for at least 30-40$ only because of them.

pretty much what i thought about, thanks for confirming (i have 2 Bitfenix Alchemy 6pin and 8pin from a older build and they are 11-16chf a piece)

 

[stulid]

The reality is Haswell testing needs to be officially certified by Intel and Fury has passed the test. That’s why we listed on spec it’s Haswell tested. As we’re not sure if this TechPower Up Haswell testing follow Intel certified standard, we’re still talking to the reviewer to see how he tested and why he has different result.

Also remember the price includes these Alchemy style braided cables, Corsair for example charge £65 on top of the price of their PSU's.

 

[crmaris]

Since I can't put a screenshot of Intel's table for the Haswell compatibility test (it is not public), I will provide this link.

http://www.jonnyguru.com/forums/showthread.php?t=10253&highlight=haswell

For Haswell compatibility you load the 12V rail with minimum load (that is 0.1A) and IF the PSU haves a second +12V rail you apply 0.05A load to it (however with the most recent ATX spec 2.4 you can apply zero load as well to the second +12V rail). At the same time you dial full load on the minor rails and the PSU has to stay within ATX spec voltage regulation limits.

Your PSU with 160 W on the minors stated max combined power clearly cannot pass this test. However as I state in my review with 100 W on the minors it was able to pass it.
Simply put, Haswell compatibility cannot be achieved (at least easily if not at all) by a group regulated PSU.

 

[evulmunk33]

Sorry for Stus comment Aris, Im sure he didnt mean to question your testing methods
our engineers specified the 5v and 3.3v rails, we didnt increase the wattage and current rating for marketing
And our engineers insist FURY passes the Intel C6/C7 test, with the current specifications...
EDIT: I just heard that we have a certificate from an independent Intel certified lab that FURY supports Intel Haswells C6 and C7 ultra low power sleep modes.
Let me hook you up with our engineers to discuss?
Theyd like to know more about your testing procedure.

Thanks a lot for your feedback and constructive criticism aris!
We are checking and are considering to adjust the 5V and 3.3V spec and fine tune the PSU
Those high current ratings dont really make sense in the real world anyhow, and only make it harder or impossible for the unit to pass some synthetic load tests.
From my understanding, by lowering the 5V and 3.3V spec and thus reducing their loading in tests, FURY should perform much better in the synthetic tests, correct?

In real world usage there is no problem, this is just a spec/labeling issue, correct?
Thanks again for the feedback! seems we tied our shoelaces together and fell flat on our face with the current labeling on FURY -_-

 

[GreiverBlade]

seems we tied our shoelasses together and fell flat on our face with the current labeling on FURY -_-

well : the look is furiously stunning ... so Fury fit wells

 

[crmaris]

with 160 W on minors during CL1 (aka Haswell test) the Fury meets trouble (red fonts mean that the measurements are out of ATX spec).

with 100 W although the 12V rail is too high (12.6V is the limit), still it is in spec while the 5V rail is doing much better (however in this rail ripple is close to 50 mV at least at the high ambient I conducted the test). So it passes it but it still is very close to the limits. Hasswell compatibility is really tough for group-regulated PSUs. Also I am aware that in real life it might have no meaning but from the moment Intel demands it I have to ran it and take its results under consideration.

 

[crmaris]

EDIT: I just heard that we have a certificate from an independent Intel certified lab that FURY supports Intel Haswells C6 and C7 ultra low power sleep modes.
Let me hook you up with our engineers to discuss?
Theyd like to know more about your testing procedure.

Sure no problem.

 

[W1zzard]

I just heard that we have a certificate from an independent Intel certified lab that FURY supports Intel Haswells

Any way you could post that certificate? Would be interesting to see how the results differ.

 

[GreiverBlade]

Why would you score an 8.3 if it fails?

well as long as it's made by FSP (OEM) I will still Buy this.

it doesn't fail everything so why give a lower score when the only fail is the Haswell certification? (not that's a totally useful certification )
overall the build quality and the sleeving get the most of it (and some other test than the Haswell Certification...)
OEM is FSP and based on the Aurum series indeed

 

[Rectifier9c]

This will definitely my next PSU. You may ask me why?

since before I got FSP PSU; Epsilon 600w and Blue storm 500w, still working until now no problem whatsoever.

 

[evulmunk33]

Any way you could post that certificate? Would be interesting to see how the results differ.

Yeah, I thought the same, im checking and will let you guys know once i got a copy!

 

[evulmunk33]

Why would you score an 8.3 if it fails? or maybe the reviewer got a faulty or defective psu (early sample).

My guess its because the test it fails is a synthetic test and its very very unlikely a pc will put 160W load on both 5v and 3.3v while there is close to 0 load on the 12V rail.
Think of the radeon 295x2, it violates the power adapter specs and draws more current than allowed with 2x8pin, but that doesnt make it a bad vga, right?
In the real world all highend PSUs can handle this load, and it wont be an issue... so its a bad thing they dont meet the spec, but the card actually works fine and performs very well.

 

[W1zzard]

My guess its because the test it fails is a synthetic test and its very very unlikely a pc will put 160W load on both 5v and 3.3v while there is close to 0 load on the 12V rail.
Think of the radeon 295x2, it violates the power adapter specs and draws more current than allowed with 2x8pin, but that doesnt make it a bad vga, right?
In the real world all highend PSUs can handle this load, and it wont be an issue... so its a bad thing they dont meet the spec, but the card actually works fine and performs very well.

But AMD doesn't claim their card is within PCIe spec. Also since the "Haswell Test" was made for Haswell, shouldn't there be a real life scenario in which it can happen? Otherwise that test wouldn't exist?

 

[evulmunk33]

But AMD doesn't claim their card is within PCIe spec. Also since the "Haswell Test" was made for Haswell, shouldn't there be a real life scenario in which it can happen? Otherwise that test wouldn't exist?

Still, breaking the spec isnt a good thing and they could have gone triple 8pin, but that doesnt make it a bad VGA, right?
Some heatsinks and mainboards break the intel spec for the cpu socket keepout zone as well, but most people never notice it and its usually highend heatsinks and mainboards which are definitely not BAD products.
Im just saying, breaking a spec by itself doesnt necessarily mean a product is bad

About why intel specified the test the way they did, im curious to know...
I worked with several PSU engineers and vendors and nobody seems to know why, and even my intel contacts dont know haha
But that doesnt matter, we guarantee that our PSU passes the Intel Haswell C6/C7 test, no matter how irrelevant to real world usage it might be, and we are sticking to it!
Theres a couple of internal emails flying back and forth right now about the C6/C7 test procedure and different test reports, and ill let you know once theres an update and I can post some official statement.

 

[W1zzard]

Still, breaking the spec isnt a good thing and they could have gone triple 8pin, but that doesnt make it a bad VGA, right?

There is no spec for anything beyond 6+8 (or 8+8, not 100% sure and too lazy to look into the spec)

Im just saying, breaking a spec by itself doesnt necessarily mean a product is bad

nobody said that

 

[evulmunk33]

nobody said that

Mhhh maybe i misunderstood rectifier9c, i thought thats what he was asking...

Why would you score an 8.3 if it fails?

btw: I received an official statement and mailed it to you and aris

 

[thubleau]

You might like to update your review.
I checked the Bitfenix site and as of September the 1st 2014 this unit has passed Intels Haswell compatability specifications.
Test results and papers are available on site.

 

[thubleau]

it doesn't fail everything so why give a lower score when the only fail is the Haswell certification? (not that's a totally useful certification )
overall the build quality and the sleeving get the most of it (and some other test than the Haswell Certification...)
OEM is FSP and based on the Aurum series indeed

so that means now that it has passed the Haswell specs you will have to change the rating from 8.3 to ????

 

[GreiverBlade]

so that means now that it has passed the Haswell specs you will have to change the rating from 8.3 to ????

nope since that certification is worth : "nothing"

 

[evulmunk33]

From what I understand the intel certification just confirms what aris concluded as well, in a real world scenario no haswell system will have an issue with this PSU, even with C6/C7 enabled.

 

[RealNeil]

Damn good looking PSU