1. what is with this phrase all the time "quietly started to sell" or "quietly added" why quietly??
2. please rephrase "powered by a cut-down version" because it is misleading. Since this article is about a single slot mini-itx compatible card, your phrase made me think that the 460 was cut down even further to make that accommodation. But it is not the case at all.
quietly started to sell" or "quietly added " usually means.. no announcement, or fan fair about the card to let the public know.. it has been released...
"powered by a cut-down version" my guess... means parts of the gpu have been disabled.like the stream processors... so it will work in the design that it is, for the heat and power aspects ...
1) Quietly added means no announcement. A quiet launch. Just appears in stores and their website.
2) Powered by a cut-down of the Polaris 11 refers to it is not a full Polaris 11 chip. The RX 460 is a cut down version of the full Polaris 11, found in the RX 480M, which normally has 1024 SPs, not the 896 that the RX 460 has.
Or they could do it like my HiS low-profile passively cooled Radeon 6570... And have the heatsink go over the top of the card and accompodate the back of the GPU as well.
As already noted by a couple of other readers, these are more-or-less standard terms for us. But that's good feedback on the 460 cut-down thing; I'll tweak that paragraph a bit.
Is there a modern gfx card that can drive a 4k display (video, no gaming) <50W? I have a HTPC with a 192W pico psu and these 75W cards are just too much. The quad core sandy bridge is sufficient, but the IGD needs to be upgraded for my 4k OLED.
The Nvidia Shield TV plays video at 4K60 and uses a 40w brick, so it's certainly possible. It's a custom Maxwell chip though. As for desktops, the GTX 750is the lowest wattage 4k-capable PCIe card, and it's rated at 55 watts. AMD offers a low profile 4K60 capable card, the R7 240, but it idles at 55 watts and ramps up to 122 watts when playing 4K video, so it's out of the picture for your needs.
Those are wall power measurements, not for the GPU as a whole. While a handful of sites have invested in the instrumented PCIe risers in the aftermath of power-gate needed to measure PCIe draw directly, most have not and that article dates to 2014; at which point in time I'm not aware of anyone doing so.
With the screen off and the system at sleep Tom's test system draws 57W; the difference between that and the GPU at full load of 122W is 65W, but those 65W also include: the power used by the CPU to feed the GPU, the power used by any system fans (case, CPU, etc) that turned back on when the system was woken up, as well as PSU efficiency losses. The oversized 850W PSU Toms tested with is only about 76% efficient running the GPUs (and considerably less at idle). The 122W wall power number only corresponds to ~95W under load. The 65W idle number is probably somewhere between 40 and 50W DC (can't find numbers, but non- 80+ Platinum PSUs are generally utterly awful at 10% load). Remembering the various other loads on the system that can'be directly quantified somewhere 25-40W looks about right for this card.
Doing the same thing with some of the higher power cards in the review indicates that they're all also hitting around their rated TDPs.
Yep, I completely misread that article. Ryan Smith below wrote that the R7 cards don't actually have a video decode block, yet they are advertised as 4K capable. So now I'm not sure even those cards would fit the requirements.
Bear in mind that the Radeon 240/250 are based on Oland, and Oland doesn't have a video decode block. The odds are your CPU has one, but it's still a potential issue.
Many of the recent highish-end SoCs can deliver 4k video on integrated graphics.
NUCs, mini-PCs, ARM-based systems like the Wetek hub/play/core, ODROID C2, Xiaomi's Mi Box, etc. Use a LibreELEC distribution (bare-bones Linux + Kodi (used to be XBMC)) and get a HTPC for less than the cost of this card.
Of course, it depends on your exact requirements ;)
I bet this will stay under 50W for 4K display. Can't say for sure but it won't go over 75W when gaming on avg and driving displays uses fixed hardware that isn't super energy expensive.
If you wanted to be extra sure you can limit its GPU clock to like 1 GHz and lower the voltage to match in Wattman. But it'll probably be under 50W without tweaks with no gaming.
Huh. I have owned one of these for a week already and had no idea they were so new. It's in my Lenovo TS440 doing PCI-E pass through in esxi 6.5 to a win10 gaming vm... I play ffxiv all day long at 2560x1600 on "laptop high" with solid fps.
I've got the 2 GiB version of that card in my home server, that also doubles as my HTPC. To be honest, it reminded me why I usually only buy Nvidia. But this was and still is the only HDMI 2.0b single-slot card available.
The issues were: - Boot freeze with my Supermicro board when using the EFI OpRom of that card. Probably because of more advanced stuff like IPMI redirect, which worked fine with a Nvidia GT730. With the legacy OpRom it then worked, but that disabled IPMI. - A terrible driver were the panel with the additional options like the important HDMI color space selection couldn't be launched. At least the card uses YgYbCr 4:4:4 at 4K/60 by itself, but I would prefer RGB. Currently the necessary setting (which existed in the past) can't simply be set. - The whole 0.5W ZeroCore thing doesn't seem to work at all anywhere, which means idle with no display output of around 4W, which is ok but not great.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
24 Comments
Back to Article
darckhart - Saturday, February 25, 2017 - link
to the author:1. what is with this phrase all the time "quietly started to sell" or "quietly added" why quietly??
2. please rephrase "powered by a cut-down version" because it is misleading. Since this article is about a single slot mini-itx compatible card, your phrase made me think that the 460 was cut down even further to make that accommodation. But it is not the case at all.
Qasar - Saturday, February 25, 2017 - link
quietly started to sell" or "quietly added " usually means.. no announcement, or fan fair about the card to let the public know.. it has been released..."powered by a cut-down version" my guess... means parts of the gpu have been disabled.like the stream processors... so it will work in the design that it is, for the heat and power aspects ...
PCTC2 - Saturday, February 25, 2017 - link
1) Quietly added means no announcement. A quiet launch. Just appears in stores and their website.2) Powered by a cut-down of the Polaris 11 refers to it is not a full Polaris 11 chip. The RX 460 is a cut down version of the full Polaris 11, found in the RX 480M, which normally has 1024 SPs, not the 896 that the RX 460 has.
StevoLincolnite - Saturday, February 25, 2017 - link
Ironically... You can unlock the shaders on the Radeon 460 so it has the full 1024 Shaders.With that in mind... This card needs to be low profile... :(
renz496 - Sunday, February 26, 2017 - link
if they go low profile they most likely have to give up single slot cooler.StevoLincolnite - Sunday, February 26, 2017 - link
Or they could do it like my HiS low-profile passively cooled Radeon 6570... And have the heatsink go over the top of the card and accompodate the back of the GPU as well.prisonerX - Saturday, February 25, 2017 - link
Another ESL victim.Gasaraki88 - Wednesday, March 1, 2017 - link
LOL, yeah but the ironic thing is he's a white dude.Ryan Smith - Sunday, February 26, 2017 - link
As already noted by a couple of other readers, these are more-or-less standard terms for us. But that's good feedback on the 460 cut-down thing; I'll tweak that paragraph a bit.FwFred - Saturday, February 25, 2017 - link
Is there a modern gfx card that can drive a 4k display (video, no gaming) <50W? I have a HTPC with a 192W pico psu and these 75W cards are just too much. The quad core sandy bridge is sufficient, but the IGD needs to be upgraded for my 4k OLED.kaidenshi - Saturday, February 25, 2017 - link
The Nvidia Shield TV plays video at 4K60 and uses a 40w brick, so it's certainly possible. It's a custom Maxwell chip though. As for desktops, the GTX 750is the lowest wattage 4k-capable PCIe card, and it's rated at 55 watts. AMD offers a low profile 4K60 capable card, the R7 240, but it idles at 55 watts and ramps up to 122 watts when playing 4K video, so it's out of the picture for your needs.PixyMisa - Sunday, February 26, 2017 - link
What? The R7 240 is a 30W card.kaidenshi - Sunday, February 26, 2017 - link
http://www.tomshardware.com/reviews/radeon-r7-240-...StevoLincolnite - Sunday, February 26, 2017 - link
No way does the Radeon R7 250 use 122 watts. It doesn't have PCI-E power... Thus it can only draw up-to 75w from the PCI-E slot.kaidenshi - Sunday, February 26, 2017 - link
According to the link I posted in reply to PixyMisa above, it can and does.DanNeely - Sunday, February 26, 2017 - link
Those are wall power measurements, not for the GPU as a whole. While a handful of sites have invested in the instrumented PCIe risers in the aftermath of power-gate needed to measure PCIe draw directly, most have not and that article dates to 2014; at which point in time I'm not aware of anyone doing so.With the screen off and the system at sleep Tom's test system draws 57W; the difference between that and the GPU at full load of 122W is 65W, but those 65W also include: the power used by the CPU to feed the GPU, the power used by any system fans (case, CPU, etc) that turned back on when the system was woken up, as well as PSU efficiency losses. The oversized 850W PSU Toms tested with is only about 76% efficient running the GPUs (and considerably less at idle). The 122W wall power number only corresponds to ~95W under load. The 65W idle number is probably somewhere between 40 and 50W DC (can't find numbers, but non- 80+ Platinum PSUs are generally utterly awful at 10% load). Remembering the various other loads on the system that can'be directly quantified somewhere 25-40W looks about right for this card.
Doing the same thing with some of the higher power cards in the review indicates that they're all also hitting around their rated TDPs.
http://www.jonnyguru.com/modules.php?name=NDReview...
kaidenshi - Sunday, February 26, 2017 - link
Yep, I completely misread that article. Ryan Smith below wrote that the R7 cards don't actually have a video decode block, yet they are advertised as 4K capable. So now I'm not sure even those cards would fit the requirements.Ryan Smith - Sunday, February 26, 2017 - link
Bear in mind that the Radeon 240/250 are based on Oland, and Oland doesn't have a video decode block. The odds are your CPU has one, but it's still a potential issue.eldakka - Sunday, February 26, 2017 - link
Many of the recent highish-end SoCs can deliver 4k video on integrated graphics.NUCs, mini-PCs, ARM-based systems like the Wetek hub/play/core, ODROID C2, Xiaomi's Mi Box, etc. Use a LibreELEC distribution (bare-bones Linux + Kodi (used to be XBMC)) and get a HTPC for less than the cost of this card.
Of course, it depends on your exact requirements ;)
kaidenshi - Sunday, February 26, 2017 - link
FwFred was specifically asking for an addon desktop GPU as he already has an HTPC.genekellyjr - Sunday, February 26, 2017 - link
I bet this will stay under 50W for 4K display. Can't say for sure but it won't go over 75W when gaming on avg and driving displays uses fixed hardware that isn't super energy expensive.If you wanted to be extra sure you can limit its GPU clock to like 1 GHz and lower the voltage to match in Wattman. But it'll probably be under 50W without tweaks with no gaming.
Ampidire - Sunday, February 26, 2017 - link
Huh.I have owned one of these for a week already and had no idea they were so new. It's in my Lenovo TS440 doing PCI-E pass through in esxi 6.5 to a win10 gaming vm... I play ffxiv all day long at 2560x1600 on "laptop high" with solid fps.
SentinelBorg - Thursday, March 2, 2017 - link
I've got the 2 GiB version of that card in my home server, that also doubles as my HTPC. To be honest, it reminded me why I usually only buy Nvidia. But this was and still is the only HDMI 2.0b single-slot card available.The issues were:
- Boot freeze with my Supermicro board when using the EFI OpRom of that card. Probably because of more advanced stuff like IPMI redirect, which worked fine with a Nvidia GT730. With the legacy OpRom it then worked, but that disabled IPMI.
- A terrible driver were the panel with the additional options like the important HDMI color space selection couldn't be launched. At least the card uses YgYbCr 4:4:4 at 4K/60 by itself, but I would prefer RGB. Currently the necessary setting (which existed in the past) can't simply be set.
- The whole 0.5W ZeroCore thing doesn't seem to work at all anywhere, which means idle with no display output of around 4W, which is ok but not great.
Itselectric - Monday, March 6, 2017 - link
They should have made the heatsink full copper, to reduce noise levels and temperatures, especially in the HTPC market that this is targeted for.