Tom's Hardware re-examines the RX 480's power consumption

[Briggsy]

24 minutes ago, xAcid9 said:

Someone that created the BIOS for 480 probably half asleep/drunk since this can be fix via vBIOS.

 

http://www.overclock.net/t/1604979/a-temporary-fix-for-the-excess-pci-e-slot-power-draw-for-the-reference-rx-480-cards/0_20

[zMeul]

6 minutes ago, airdeano said:

your first graphic showed the 3.3 passing through the PCIe slot connector, then your follow up sows that same 3.3 powered through the PEG (PSU lead) ("PEG" stands for PCI Express Graphics)

you are confusing some shit

PEG is the PCIe slot - PCI Express Graphics, PEG

the power connector is the PCIe 6 pin

 

the 1st picture is outdated / wrong - Tom's removed it

since then, they made manual measurements to determine how the phases link, @xAcid9 posted the updated pic

 

ps: next time, please quote me if you want me to see your post, I might miss it

[airdeano]

me thinks that is incorrect (in bold and underlined passage):

 

Quote

 

6 pin PCI Express power cable

This cable is used to provide extra 12 volt power to PCI Express expansion cards. PCI Express motherboard slots can provide a maximum of 75 watts. Many video cards draw significantly more than 75 watts so the 6 pin PCI Express power cable was created. These high-power cards draw most of their power from the 12 volt rail so this cable provides only 12 volts. These are sometimes called "PCI Express cables". They are also occasionally called "PEG cables" where "PEG" stands for PCI Express Graphics. If your power supply doesn't have a 6 pin PCI Express cable then you can use the adapter shown above on the right to convert two 4 pin peripheral cables into a PCI Express cable. If you use an adapter then be sure to plug the 4 pin peripheral connectors into separate cables coming from the power supply. If you plug them both into the same power supply cable then you are drawing all the power of the PCI Express connector through a single 18 gauge wire. You can usually get away with that but there's no reason to do it. The PCI Express 6 pin connector is polarized so it can only be plugged in pointing in the correct direction. But as with connectors of this type, you can sometimes force them into the wrong kind of socket if you try hard enough. If it doesn't slide in easily then you're probably plugging it into the wrong place.

Some video cards come with the 8 pin PCI Express power connector to support higher wattage than the 6 pin PCI Express connectors. It's okay to plug a 6 pin PCI Express power cable into an 8 pin PCI Express connector. It's designed to work that way but will be limited to the lower wattage provided by the 6 pin version of the cable. The 6 pin cable only fits into one end of the 8 pin connector so you can't insert it incorrectly but you can sometimes force the 6 pin cable in the wrong way if you try hard enough. Video cards can sense whether you have plugged a 6 pin or 8 pin cable into an 8 pin connector so the video card can impose some kind of restriction when running with only a 6 pin power cable. Some cards will refuse to run with only a 6 pin cable in an 8 pin socket. Others will work with a 6 pin cable at normal speeds but will not allow overclocking. Check the video card documentation to get the rules. But if you don't have any other information then just assume that if your video card has an 8 pin connector then you must plug in an 8 pin cable.

source: http://www.playtool.com/pages/psuconnectors/connectors.html (since 2008)

 

the PCIe is a slot on the motherboard. Peripheral Component Interconnect Express (PCIe or PCI-E) is a serial expansion bus standard for connecting a computer to one or more peripheral devices  the PEG is the PSU lead.

 

[NumLock21]

PEG is for the 6-pin and 8-pin power connectors that's on the graphic cards. Most will just say power connector, so it's easier for non techies and wider range of audiences to know what that person is talking about. The proper name for them is PEG. I can say the GTX 1080 uses a single 6+2 pin PEG and a lot will go wat??!!!.

And proper name for SATA III is SATA 6Gb/s (SATA 6G) not SATA III.

[zMeul]

12 minutes ago, airdeano said:

the PCIe is a slot on the motherboard. Peripheral Component Interconnect Express (PCIe or PCI-E) is a serial expansion bus standard for connecting a computer to one or more peripheral devices  the PEG is the PSU lead.

+ @NumLock21 NO!  what the hell, who told you that!?

just go into the BIOS of any motherboard with a PCIe 16x that is used for graphics, it's referred to as PEG - PEG is the PCIe 16x slot that is specifically designated to work with graphic cards; the PEG is the 1st slot where the BIOS searches for the video card

[airdeano]

@zMeul PCIe also is used for x8, x4, x1 devices and not restricted to just graphic cards.. there are standards and this is a standard, not motherboard nomenclature.

[zMeul]

4 minutes ago, airdeano said:

@zMeul PCIe also is used for x8, x4, x1 devices and not restricted to just graphic cards.. there are standards and this is a standard, not motherboard nomenclature.

some PCIe slots (8x, 4x, 1x) cannot accept 16x length boards - they cannot be physically inserted into

 

ignoring the battery and the heat sink, the 1X and 4X slots are physically sealed at the end making it impossible to insert a 16X card in them

[airdeano]

you were talking about the 1st x16 slot.. being for graphics only.. 

 

@zMeul i went to: https://pcisig.com/specifications/ and there is no mention of PEG as a relatable term (https://pcisig.com/search/node/PEG), use Peripheral Component Interconnect Express and i get this: https://pcisig.com/search/node/Peripheral Component Interconnect Express

[zMeul]

11 minutes ago, airdeano said:

you were talking about the 1st x16 slot.. being for graphics only.. 

 

@zMeul i went to: https://pcisig.com/specifications/ and there is no mention of PEG as a relatable term (https://pcisig.com/search/node/PEG), use Peripheral Component Interconnect Express and i get this: https://pcisig.com/search/node/Peripheral Component Interconnect Express

the PEG term is used by motherboard manufacturers (I think ASUS used it 1st about 12y ago) to determine the location of the graphics card as it's possible to disable it, run on integrated graphics or reconfigure it

and I didn't said it's for graphics only! the 1st slot is where BIOS starts searching for the video card - the 1st 16x slot is referred to as PEG

as i said, go into a mobo's BIOS and look for it

[airdeano]

my training since 2001 was PEG is PSU power lead and PCIe is a slot on the motherboard. and sense PCI-SIG is the spec.. i'll use its nomenclature.

also the slot is not god-like and can be modified for oversized cards to accommodate the interface issue.

 

 

[NumLock21]

25 minutes ago, zMeul said:

+ @NumLock21 NO!  what the hell, who told you that!?

just go into the BIOS of any motherboard with a PCIe 16x that is used for graphics, it's referred to as PEG - PEG is the PCIe 16x slot that is specifically designated to work with graphic cards; the PEG is the 1st slot where the BIOS searches for the video card

They got it wrong. PEG is for the power connector, not the physical slots on the motherboard. Just Google image 6 pin PEG or 8 pin PEG and look at what comes up.

[zMeul]

17 minutes ago, airdeano said:

also the slot is not god-like and can be modified for oversized cards to accommodate the interface issue.

mate, why are we talking about mods?! when I was talking about physically inserting 16x cards into 1x, 4x, 8x slots?!

here, a 1x slot opened at the end that can accept higher length cards:

4x and 8x open ended:

Quote

my training since 2001 was PEG is PSU power lead and PCIe is a slot on the motherboard. and sense PCI-SIG is the spec.. i'll use its nomenclature.

you were unable to find PEG in PCI-SIG white papers, the term is not PCI-SIG standard and whomever instructed you, was wrong - I guess you can use the term "PEG power connector" as you can use the term "PCIe 6/8 pin power connector"; but, "PEG" refers to the 16x slot

 

@NumLock21 as I said, it's the wrong terminology, and it's not standard terminology either

don't quote me on it, but I think ASUS has a patent on PEG link speed - and it's specifically referring to the PCIe 16x slot where the video card is

 

---

 

why is it referred to as PEG? because in most cases where the CPU has only 16 PCIe lanes, all of those link to the 1st PCIe 16X slot

 

---

 

I'm done debating this further

[PerfectTemplar]

15 hours ago, Valentyn said:

I like how they mention dirty components. The one person so far to claim it blew this motherboard had a system that looked like this.

 

The system is in a terrible state

I just wanted to point out that that PSU looks like the more worrying component when it comes to power draw issues.

[HalGameGuru]

I can't say this guy is right and others are wrong but he breaks down the power delivery a bit differently. Making reference to similarities and differences with the earlier Fury cards.

 

Aside from that, if his explanation is correct, this should be fixable via the VRM controller, which is apparently a top of the line part. Apparently a lot of the PCB is pretty straight forward and as expected from AMD cards of the past.

[HalGameGuru]

10 minutes ago, PerfectTemplar said:

I just wanted to point out that that PSU looks like the more worrying component when it comes to power draw issues.

I'm thinking, with the robustness of the motherboard hardware involved the bigger issue may be ripple and cheap PSUs. I look forward to more info

 

As an aside there is going to be a lot of interesting info coming out as the hardware is more thoroughly inspected.

[xAcid9]

18 minutes ago, HalGameGuru said:

I can't say this guy is right and others are wrong but he breaks down the power delivery a bit differently. Making reference to similarities and differences with the earlier Fury cards.

-snip-

Aside from that, if his explanation is correct, this should be fixable via the VRM controller, which is apparently a top of the line part. Apparently a lot of the PCB is pretty straight forward and as expected from AMD cards of the past.

Yup, The Stilt already did that. It did not fix the 3v3 phases in RX-480 but it does re-balance the load distribution between PCIE slot and the 6 pin. 

According to Ryan Shrout, Allyn from PCPer tested The Stilt's fix and appear to be "working". http://www.overclock.net/t/1604979/a-temporary-fix-for-the-excess-pci-e-slot-power-draw-for-the-reference-rx-480-cards/0_20

[Sintezza]

10 hours ago, xAcid9 said:

They updated the picture. 

 

Similar to what buildzoid found in his 2nd video.

 

 

 

 

Whats wrong with their first graphs?

 

Is that a updated picture from Toms?

Lol seriously, this makes me laugh and facepalm pretty hard.

They claim to be hardware and vrm cirquitry experts?

 

This is realy not worth my time.

 

lmao

 

 

[Briggsy]

Not sure why the PEG debate is even a thing, it seems like a whole lot of semantics. Years ago it was referred to as PEG slot and PEG connector by some, Tom's hardware and PCper don't refer to the slot as PEG in their written 480 reviews or power testing follow-ups. Tom's calls it the mainboard slot, PCper calls its the PCI Express Slot.

 

Guess there will be another round of testing in the next few days anyway.

[The Benjamins]

22 minutes ago, Sintezza said:

 

Is that a updated picture from Toms?

Lol seriously, this makes me laugh and facepalm pretty hard.

They claim to be hardware and vrm cirquitry experts?

 

This is realy not worth my time.

 

lmao

 

 

No, they updated the picture base on information from the community. someone said they traced out were the power is coming from that that's what they got.

[Sintezza]

34 minutes ago, The Benjamins said:

No, they updated the picture base on information from the community. someone said they traced out were the power is coming from that that's what they got.

They should educate them selfs a bit better in vrm cirquitry i think.

Currious how manny diffrent scheme´s we still going to see.

I´m going to wait untill we see original and true scheme´s of how the vrm cirquitries are exaly build up on this card.

Because this is realy not worth my time. .

 

AMD made a statement, and broaght a fix for it.

So that is that.

 

I personaly allready have a pretty clear view on how the vrm cirquits for both the VCCGT and the Vram are "most likely" build up and routed.

But i´m not going to trow that in the wild, untill i have seen the original scheme´s that would proof my sight..

Or wenn i have the possibility to look a the pcb myself.

But all i can say right now, is that the media makes one big mess of this.

And blowing it totaly out of porportion.

 

[Doobeedoo]

Lol that picture. Though hopefully driver rectifies the power draw issue.

[Briggsy]

So if they can lower the power draw on the slot through software, what's to stop someone from doing the opposite, like a virus, or even a mistake made in a driver update?

 

It's easy enough to downplay how serious that is, but weirder things have happened in driver updates.

[Mira Yurizaki]

36 minutes ago, Briggsy said:

So if they can lower the power draw on the slot through software, what's to stop someone from doing the opposite, like a virus, or even a mistake made in a driver update?

 

It's easy enough to downplay how serious that is, but weirder things have happened in driver updates.

This is what Furmark does (and for some reason why people think it's the dumbest thing to run)

[Coaxialgamer]

23 hours ago, AresKrieger said:

Well to be fair many people who buy budget hardware buy cheap components, or are still using old hardware. Honestly it's only a potential issue to better parts if you take it off stock settings, however it doesn't change the fact the reference model 480 is crap, and should not be recommended when better versions will be out soon enough, also it can still cause your system to crash with better hardware just for reference.

Pretty sure they didn't mean "budget"  parts.  Rather the "dont-touch-this-with-a-ten-feet-poll"  cheap.  A great example of that would be logysis power supplies,  for example.  

[xAcid9]

1 hour ago, Briggsy said:

So if they can lower the power draw on the slot through software, what's to stop someone from doing the opposite, like a virus, or even a mistake made in a driver update?

 

It's easy enough to downplay how serious that is, but weirder things have happened in driver updates.

You mean like 196.75 and 320.18? 

Anyway, AMD have been using that same controller since 7000 series, how many driver burning update they released since then?