Question Molex to PCIe limit.

Jul 14, 2019
14
0
10
0
How many watts of consumption from the GPU is too much when using a Molex to PCIe adaptor that would be unsafe? If I have a 600 watts 80 plus bronze PSU with only one 6pin pci-e connector and I want to use a MSI GTX 670 OC (requires 2x6 pin connectors), would it be safe? Or would the needs of the GPU be too much for the PSU to handle through Molex? I've read somewhere that Molex can give 40 watts, so if the PCI-E x16 3.0 slot can give 150 watts (limited to 75 by the GPU manufacturer), and the 6pin pci-e connector gives 75 watts, would it be safe to try to get the missing watts from the molex to PCI-e adaptor?

And does the GPU draw only the needed amount of watts from the molex or it takes the whole 40 watts from it? In case I'm not wrong and they actually provide 40 watts, either way, would it take only how much it needs or would I risk frying the GPU when using Molex to PCI?

P.S. I've read in some pages that the GTX 670 OC uses 170 watts, but on some other pages I've read that it uses 190, and on other pages I've read it uses over 200 watts, so... How much does it really use?
P.S.2. I'm pretty sure that my GPU is the MSI GTX 670 Power Edition, not OC.
P.S.3. If I just underclock it would it be enough?
I know it's pretty messy, I'm sorry D:
 
Last edited:

John Chesterfield

Reputable
Mar 22, 2015
34
6
4,545
1
Well... The GPU comes with the adaptor for a reason.
Yes, the reason is to make the GPU more sellable, because it includes an accessory that people think they then don't need to purchase extra. The fact that it's electrically not sensible doesn't matter to the GPU manufacturer because once the GPU is sold, if one of those adapters fails and the PSU blows, people will blame the power supply and not the GPU. Moreover, more chance the power supply fails, because use of such an adapter means that the PSU wasn't designed to run such a high power graphics card. So might as well chuck one in for a coupla bucks.

You've had good advice here, you can choose to ignore it if you like and hook everything up. It might work, might work for quite a while.

Would I trust such a set up? Not a chance.
 
Reactions: LauroSalvatore

AllanGH

Commendable
Mar 10, 2019
2,109
417
1,740
62
If I have a 600 watts 80 plus bronze PSU with only one 6pin pci-e connector and I want to use a MSI GTX 670 OC (requires 2x6 pin connectors), would it be safe?
The short answer is....NO.

You are free, however, to do anything you want--unsafe or not. Just do not expect people to tell you that it is safe to do something that is NOT SAFE.

If your PSU does not have the required connectors, you need a new PSU.
Nothing detracts from Ret's good advice.
 
Reactions: LauroSalvatore
How many watts of consumption from the GPU is too much when using a Molex to PCIe adaptor that would be unsafe? If I have a 600 watts 80 plus bronze PSU with only one 6pin pci-e connector and I want to use a MSI GTX 670 OC (requires 2x6 pin connectors), would it be safe? Or would the needs of the GPU be too much for the PSU to handle through Molex? I've read somewhere that Molex can give 40 watts, so if the PCI-E x16 3.0 slot can give 150 watts (limited to 75 by the GPU manufacturer), and the 6pin pci-e connector gives 75 watts, would it be safe to try to get the missing watts from the molex to PCI-e adaptor?

And does the GPU draw only the needed amount of watts from the molex or it takes the whole 40 watts from it? In case I'm not wrong and they actually provide 40 watts, either way, would it take only how much it needs or would I risk frying the GPU when using Molex to PCI?

P.S. I've read in some pages that the GTX 670 OC uses 170 watts, but on some other pages I've read that it uses 190, and on other pages I've read it uses over 200 watts, so... How much does it really use?
P.S.2. I'm pretty sure that my GPU is the MSI GTX 670 Power Edition, not OC.
P.S.3. If I just underclock it would it be enough?
I know it's pretty messy, I'm sorry D:

That would be ZERO, never use adaptors period for any reason.

If the PSU doesn't have the correct connections then you need a new PSU.
 
Reactions: LauroSalvatore
To start, that PSU isn't very efficient (80+ white) and is from an unknown brand.
The fan is sleeve bearing and very prone to failure.

Read more: https://www.jonnyguru.com/blog/2018/09/17/riotoro-builder-500w-power-supply/6/

While this PSU is good for the money and isnt terrible, it's not good by any stretch. Its only on par with an EVGA BV, which is another extremely budget unit.
The 670 is a thirsty card and would require a pretty good quality unit, which the Riototo builder isn't.
 
Jul 14, 2019
14
0
10
0
To start, that PSU isn't very efficient (80+ white) and is from an unknown brand.
The fan is sleeve bearing and very prone to failure.

Read more: https://www.jonnyguru.com/blog/2018/09/17/riotoro-builder-500w-power-supply/6/

While this PSU is good for the money and isnt terrible, it's not good by any stretch. Its only on par with an EVGA BV, which is another extremely budget unit.
The 670 is a thirsty card and would require a pretty good quality unit, which the Riototo builder isn't.
So... First of all, it is indeed efficient, at 230v it's 85% efficient, where I live my voltage is 230v. Second of all, you're right, it's a quite unknown brand, that doesn't mean it's a bad branding. Every branding starts somewhere, it's pretty dumb to judge something because it's relatively "unknown". Unless the branding has some rumors or people saying that it's actually bad; this isn't the case.
Third of all(?): Yes, the fan is indeed sleeve bearing, but I've had plenty of PSU's with that kind of fan and lasting over 3 years easily, so it's not really a problem, worst case sceneario a replace the fan. Surely it won't be over 40 dollars.


And btw, you just sent me a review where the final score is 8.2, if your point was to try and let me know that the PSU isn't good; it didn't quite work my dude.

Oh, forgot to mention, I have G4560, it runs at 54 watts, the power consumption of the GTX is between 170 and 200 watts, the PSU power is of 600watts, capping at max resources usage, it would be MAX 300watts of consumption between GPU+CPU, and you're telling me that the PSU isn't good enough?
 
it's pretty dumb to judge something because it's relatively "unknown"
No it's not. Your power supply is the backbone of your pc.
Without knowing much about the brand it's hard to know if this psu would be amazing or burn your house down. Even if everything looks ok, it's hard to tell just where they might have cut corners. Its best not to risk it with an unknown brand and stick to something you know will work.

Sure the review has a decent score, but look at the score of a known good model from Seasonic: https://www.jonnyguru.com/blog/2017/08/21/seasonic-focus-plus-550-gold-power-supply/6/
That 8.5 of the 600w doesnt seem so good when it gets beat by the 9.8 of a 550w. Yes the focus plus is more expensive, but if you spend more on a psu it will be quieter and much more reliable.

The Riotoro Builder series isn't their best. It wouldn't be what Id suggest for a 670 since Riotoro has better models available.

What country are you in? A Seasonic S12II is a tried and true unit available most everywhere.
 
Last edited:
Jul 14, 2019
14
0
10
0
No it's not. Your power supply is the backbone of your pc.
Without knowing much about the brand it's hard to know if this psu would be amazing or burn your house down. Even if everything looks ok, it's hard to tell just where they might have cut corners. Its best not to risk it with an unknown brand and stick to something you know will work.

Sure the review has a decent score, but look at the score of a known good model from Seasonic: https://www.jonnyguru.com/blog/2017/08/21/seasonic-focus-plus-550-gold-power-supply/6/
That 8.5 of the 600w doesnt seem so good when it gets beat by the 9.8 of a 550w. Yes the focus plus is more expensive, but if you spend more on a psu it will be quieter and much more reliable.

The Riotoro Builder series isn't their best. It wouldn't be what Id suggest for a 670 since Riotoro has better models available.

What country are you in? A Seasonic S12II is a tried and true unit available most everywhere.
Well... Is this one good enough? https://www.coolermaster.com/catalog/power-supplies/mwe-series/mwe-bronze-550/
 
So many overblown answers here...

While the sentiment is not on the wrong side, technically it is.

Using adapters is not "evil" or "terrible" as long as you know what your're doing. The OP is openly laying out his context and asking for help.

As for the answer itself, when talking about old PSUs, some only come with a single PCIe power molex (6/8 pin; usually red) and you need to get the secondary one from a regular 4pin "fat" molex using the adapter some GPUs come with. Again, just to be super extra clear, THIS IS NOT EVIL NOR A BAD THING FOR THE PSU. All the 12v rails come from the same "source" in the PSU, be it multi-rail or single-rail (it's more obvious for these), so using the adapter is just overcoming a dumb shortfall of the design of the PSU and not forcing it to run out of spec.

So, to put it simply for this scenario, what matters are 3 things:
1.- Is the rail you'd be using the molex adapter have the necessary Amp capacity for the GPU?
2.- Is the combined 12v rail wattage not overloaded by using the adapter?
3.- Do you have enough un-used 4pin molex'es rails so they become exclusive for the GPU?

Cheers!
 

AllanGH

Commendable
Mar 10, 2019
2,109
417
1,740
62
Is the combined 12v rail wattage not overloaded by using the adapter?
This is an extremely short-sighted view of the situation

The issue is one of the overall capacity of the PSU, as well as the current (I) distribution through the 12V conductors, themselves. The use of HDD "Molex" (in actuality, AMP) connectors reduces the distribution of the current flow to a single 12V conductor, where the aggregate current flow WILL over capacity the single conductor involved in the supply circuit. The problem is hardly ameliorated when dealing with 4-pin "CPU-styled" connectors, having 2 -12V conductors; where the maximum current spec for the conductors WILL be exceeded.

The same ampacity deficit is also in play where any other adapter is used when a smaller number of supply conductors are being 'expanded' to a larger number of distribution conductors.


Using adapters is not "evil" or "terrible" as long as you know what your're doing.
The problem lies in the fact that the overwhelming majority of people who poke-around inside computers, thinking "Hee-yuck! Dis is easy. I can do dis, too!" haven't the slightest education in either CIS or EE and, in fact, have no clue what they are actually doing--they only know what they want to do, whether or not it happens to be safe, or ill-advised.

As a consequence, such individuals risk over-loading a PSU, or creating a conductor over-ampacity condition, in which PSU failure or fire are likely possible outcomes; because these individuals never stop to consider the total current flow in the conductors from the PSU to the adapter, itself, and THIS is the segment of the wiring in which overheat and fire hazards exist.

To further the recommendation that un-trained individuals implement the use of PSU harness adapters is the epitome of irresponsibility--particularly in this context, where assistance is being rendered at distance to individuals who lack an education in the applicable engineering fields.

The most ethical recommendation, in such situations is the complete avoidance of all harness adapters, and to use the correct PSU for the loads involved.
 
This is an extremely short-sighted view of the situation

The issue is one of the overall capacity of the PSU, as well as the current (I) distribution through the 12V conductors, themselves. The use of HDD "Molex" (in actuality, AMP) connectors reduces the distribution of the current flow to a single 12V conductor, where the aggregate current flow WILL over capacity the single conductor involved in the supply circuit. The problem is hardly ameliorated when dealing with 4-pin "CPU-styled" connectors, having 2 -12V conductors; where the maximum current spec for the conductors WILL be exceeded.

The same ampacity deficit is also in play where any other adapter is used when a smaller number of supply conductors are being 'expanded' to a larger number of distribution conductors.
Isn't that EXACTLY what I was careful to point out? =/

Picking out messages like this hide the full picture and allows you to make mistakes like that.

If you don't know what I mean by "rails", I precisely mean the Amps the PSU allocates to each 12V... Well... Rail...

We know the 6pin is 75W (8pin is 150W) and at 12V that means it's going to draw, within spec, up to 6.25A (8pin is 12.5A) of electricity. The math is not hard in this case and you can safely use an adapter as long as the PSU lets you know what the Wattage is for each 12v rail. Additionally, is why I pointed out why using a standalone cable/rail for the GPU is a good idea, but you did not read or understood that.

The problem lies in the fact that the overwhelming majority of people who poke-around inside computers, thinking "Hee-yuck! Dis is easy. I can do dis, too!" haven't the slightest education in either CIS or EE and, in fact, have no clue what they are actually doing--they only know what they want to do, whether or not it happens to be safe, or ill-advised.

As a consequence, such individuals risk over-loading a PSU, or creating a conductor over-ampacity condition, in which PSU failure or fire are likely possible outcomes; because these individuals never stop to consider the total current flow in the conductors from the PSU to the adapter, itself, and THIS is the segment of the wiring in which overheat and fire hazards exist.

To further the recommendation that un-trained individuals implement the use of PSU harness adapters is the epitome of irresponsibility--particularly in this context, where assistance is being rendered at distance to individuals who lack an education in the applicable engineering fields.

The most ethical recommendation, in such situations is the complete avoidance of all harness adapters, and to use the correct PSU for the loads involved.
While you're not wrong, the OP is not one of them as he has done the right thing and come to us for advice.

If there's no danger to him using an adapter, then recommending him spending money on a new PSU is irresponsible as well, don't you think? This is akin to teling a person they need to change the whole engine, because they shouldn't replace the distributor belt with a chain (or vice-versa). There's always important conditions to meet before jumping the gun.

Cheers!
 

AllanGH

Commendable
Mar 10, 2019
2,109
417
1,740
62
The "rail" is the output from each OCP, not the individual conductor.

A single rail can, and does, have multiple distribution conductors, with each conductor limited to 20A/240W.

If there's no danger to him using an adapter, then recommending him spending money on a new PSU is irresponsible as well, don't you think?
Firstly, without you being there, with the OP, you cannot positively know that there is no danger in following such a recommendation.

Secondly, you cannot trust the description, rendered by an individual, particularly where the lack of knowledge, and potential barriers of language translation are involved.

Thirdly, when the risks of adverse outcome are weighed between two alternative recommendations, it is far more responsible to select the course of action that considers the safety of the individual to which such advice is being rendered, rather than the possibility that that individual might spend an extra 20-Quid.

Your recommendation of it being OK, as long as you know it's OK to do so, particularly when common sense tells you that that conclusion cannot be competently reached by the receiver of such a recommendation, is irresponsible in the extreme. Furthermore, I have been alive and working in the field long enough to see more than my fair share of situations when such advice resulted in a decidedly "NOT OK" outcome.

"Fan-Out" adapters should be used after only the specific implementation has been fully analyzed, and the use of such an adapter has been deemed to be advisable under all loading conditions. When rendering advice from afar, that means they should never be used.

Consideration for the OP's safety, as well as the dictates of best practices, require nothing less than the recommendation to replace the PSU with a unit that is designed to handle the proposed load, as well as the safe distribution of the load currents.

[EDIT]
Please refer to: IEC 60950-1
 
Last edited:
The "rail" is the output from each regulator, not the individual conductor.

A single rail can, and does, have multiple distribution conductors.
Yes, and it's up to each individual line the max amount of power it can transit through so it doesn't snap or melt. Even for really cheap PSUs, the cables are rated WELL above the amount of power they'll have running through, so it is safe to assume they'll withstand an adapter meant to suck 75W from them.

Gosh, I can't remember where I read it, but 100% of the cables inside your PC are rated for like 10X the amount of rated power that would transit at any given time. Yes, 100% of them. Why? Because that's just the best type of cable from a cost effective perspective due to defect rates.

Firstly, without you being there, with the OP, you cannot positively know that there is no danger in following such a recommendation.

Secondly, you cannot trust the description, rendered by an individual, particularly where the lack of knowledge, and potential barriers of language translation are involved.

Thirdly, when the risks of adverse outcome are weighed between two alternative recommendations, it is far more responsible to select the course of action that considers the safety of the individual to which such advice is being rendered, rather than the possibility that that individual might spend an extra 20-Quid.

Your recommendation of it being OK, as long as you know it's OK to do so, particularly when common sense tells you that that conclusion cannot be competently reached by the receiver of such a recommendation, is irresponsible in the extreme. Furthermore, I have been alive and working in the field long enough to see more than my fair share of situations when such advice resulted in a decidedly "NOT OK" outcome.

"Fan-Out" adapters should be used after only the specific implementation has been fully analyzed, and the use of such an adapter has been deemed to be advisable under all loading conditions. When rendering advice from afar, that means they should never be used.

Consideration for the OP's safety, as well as the dictates of best practices, require nothing less than the recommendation to replace the PSU with a unit that is designed to handle the proposed load, as well as the safe distribution of the load currents.
I've burned PSUs for fun in order to know what works and what doesn't, so please don't diminish any other person's experience. I've also designed circuit boards (well, back in Uni; meh) and know perfectly well how PSUs are rated and built. I don't fix or repair electrical things for a living, so I'm not fully qualified to diagnose to a level of detail that would leave zero room for doubt, but I wouldn't be telling someone to burn their house with this sort of thing if I knew there was a big enough risk associated to it! Sometimes you need to give people the benefit of the doubt here...

I will die on this hill if I must, as I know what I'm talking about as well as you are. And you probably know that I'm not wrong, even when you're trying to leave zero room of potential failures, I also know the risk is so small and insignificant, the OP would be better off saving that money and buying a new SSD or something.

By now, the OP should have enough information in order to follow either's advice. And just to reiterate, I don't think all of the given advice about a new PSU is bad per se, but it's not the best or optimal solution given how the OP is open about knowing this topic. That's why I classify it as overblown. I mean, it's a 600W rated PSU. The chances* it has and can deliver the 75W through the regular "HDD" rail/line is not a massive caveat'ed assumption.

So, I'll stop here and let the OP chime in.

Cheers!
 

AllanGH

Commendable
Mar 10, 2019
2,109
417
1,740
62
I, long ago, got bored with the useless pursuit of slamming the cajones of responsibility against the cajones of license; and you obviously have an emotional dog in this disagreement, and will accept nothing less than a full retreat and surrender.

I wish you and your pride well.
 

DSzymborski

Illustrious
Moderator
Yes, every brand starts somewhere. And the ones that start by using top-notch manufacturing for their parts do well. Power supplies are actually manufactured by relatively few companies and there are no "secret" ones out there. Riotoro has better power supplies than this and while this power supply isn't garbage, it's also a budget power supply designed for specific power needs. Running a 670, I'd want something more robust; this would be fine if you were running a 1050 Ti.
 
Jul 14, 2019
14
0
10
0
Well then, I've got plenty of answers so I have quite more knowledge over this particular topic. And due to the lack of answers over my last question I'm guessing the CoolerMaster MWE 550W isn't good enough either. So I'm gonna take in mind Yuka's proposal and Allan's risks. First of all I'm gonna ask one last question while giving more info about my PSU. Actually, I'll post an image of it: View: https://imgur.com/3IUkDjb

With these specifications in mind, would it be able to handle the adaptor from Molex to PCI-e?

When it comes to Molex connectors, I have both of them free, none are being used, and they are in the same cable.

In case it doesn't work. Or it simply is a really bad idea, I'll just go for the CoolerMaster MWE 550. Guys, I'm not rich nor I have won the lottery. lol
 

ASK THE COMMUNITY