AMD's Trinity : An HTPC Perspective
by Ganesh T S on September 27, 2012 11:00 AM ESTIn the last few HTPC reviews, we have incorporated video decoding and rendering benchmarks. The Ivy Bridge review carried a table of values with the CPU and GPU usage. The Vision 3D 252B review made use of HWInfo's sensor graphs to provide a better perspective. In the latter review, it was easier to visualize the extent of stress that a particular video decode + render combination gave to the system. Unfortunately, HWInfo doesn't play well with the A10-5800K / Radeon HD 7660D yet. In particular, GPU loading and CPU package power aren't available for AMD-based systems yet.
The tables below present the results of running our HTPC rendering benchmark samples through various decoder and renderer combinations. Entries in bold with a single star indicate that there were dropped frames as per the renderer status reports in the quiescent state, while double stars indicate that the number of dropped frames made the video unwatchable. The recorded values include the GPU loading and power consumed by the system at the wall. An important point to note here is that the system was set to optimized defaults in the BIOS (GPU at 800 MHz, DRAM at 1600 MHz and CPU cores at 3800 MHz).
madVR :
madVR was configured with the settings mentioned in the software setup page. All the video post processing options in the Catalyst Control Center were disabled except for deinterlacing and pulldown detection. In our first pass, we used a pure software decoder (avcodec / wmv9 dmo, through LAV Video Decoder) to supply madVR with the decoded frames.
LAV Video Decoder Software Fallback + madVR | ||
Stream | GPU Usage % | Power Consumption |
480i60 MPEG-2 | 38 | 77.9 W |
576i50 H.264 | 24 | 68.2 W |
720p60 H.264 | 49 | 106.6 W |
1080i60 H.264 | 81 | 128.1 W |
1080i60 MPEG-2 | 85 | 115.4 W |
1080i60 VC-1 | 84 | 131.7 W |
1080p60 H.264 | 51 | 116.6 W |
madVR takes up more than 80% of the resources when processing 60 fps interlaced material. The software decode penalty is reflected in the power consumed at the wall, with the 1080i60 VC-1 stream consuming more than 130W on an average. The good news is that all the streams played without any dropped frames with the optimized default settings.
The holy grail of HTPCs, in our opinion, is to obtain hardware accelerated decode for as many formats as possible. A year or so back, it wasn't possible to use any hardware decoders with the madVR renderer. Thanks to Hendrik Leppkes's LAV Filters, we now have a DXVA2 Copy-Back (DXVA2CB) decoder which enables usage of DXVA2 acceleration with madVR. The table below presents the results using DXVA2CB and madVR.
LAV Video Decoder DXVA2 Copy-Back + madVR | ||
Stream | GPU Usage % | Power Consumption |
480i60 MPEG-2 | 44 | 76.8 W |
576i50 H.264 | 24 | 66.2 W |
720p60 H.264 | 54 | 102.4 W |
1080i60 H.264 ** | 72 | 111.1 W |
1080i60 MPEG-2 * | 82 | 111.8 W |
1080i60 VC-1 * | 84 | 111.6 W |
1080p60 H.264 ** | 64 | 110.4 W |
There is a slight improvement in power consumption for the first few streams. We still have a bit of power penalty compared to pure hardware decode because the decoded frames have to get back to the system memory and then go back into the GPU for madVR to process. An unfortunate point to note here is that none of the 1080i60 / 1080p60 streams could play properly with our optimized default settings (rendering their GPU usage and power consumption values meaningless). We did boost up the memory speeds to DDR3-2133 and saw some improvements with respect to the number of dropped frames. However, we were unable to make the four streams play perfectly even with non-default settings.
EVR-CP :
For non-madVR renderers, we set Catalyst 12.8 to the default settings. The table below presents the results obtained with LAV Video Decoder set to DXVA2 Native mode. All the streams played perfectly, but the power numbers left us puzzled.
LAV Video Decoder DXVA2 Native + EVR-CP | ||
Stream | GPU Usage % | Power Consumption |
480i60 MPEG-2 | 26 | 78.1 W |
576i50 H.264 | 22 | 78.1 W |
720p60 H.264 | 38 | 90.1 W |
1080i60 H.264 | 69 | 103.9 W |
1080i60 MPEG-2 | 69 | 102.2 W |
1080i60 VC-1 | 69 | 104.2 W |
1080p60 H.264 | 60 | 98.4 W |
For SD streams, the power consumed is almost as much as madVR with software decode. However, the HD streams pull back the numbers a little. This is something worth investigating, but outside the scope of this article. However, we wanted to dig a bit into this, and decided to repeat the tests with the EVR renderer.
EVR :
With Catalyst 12.8 in default settings and LAV Video Decoder set to DXVA2 Native mode, all the streams played perfectly with low power consumption. All post processing steps were also visible (as enabled in the drivers)
LAV Video Decoder DXVA2 Native + EVR | ||
Stream | GPU Usage % | Power Consumption |
480i60 MPEG-2 | 27 | 60.6 W |
576i50 H.264 | 25 | 60.1 W |
720p60 H.264 | 35 | 65.7 W |
1080i60 H.264 | 67 | 80.1 W |
1080i60 MPEG-2 | 67 | 80.6 W |
1080i60 VC-1 | 67 | 82.5 W |
1080p60 H.264 | 59 | 79.2 W |
A look at the above table indicates that hardware decode with the right renderer can make for a really power efficient HTPC. In some cases, we have more than 20 W difference depending on the renderer used, and as much as 40 W difference between software and hardware decode with additional renderer steps.
49 Comments
View All Comments
Beenthere - Thursday, September 27, 2012 - link
There is an appropriate CPU/APU model for every budget these days. Virtually any current model APU/CPU will perform just fine for 98% of consumers. Most consumers buy what fits their needs and budget, not the over priced, over hyped top-of-the-line models.AMDs new Trinity APUs and Vishera desktop FX processors offer more performance for less, which is good for consumers.
silverblue - Thursday, September 27, 2012 - link
We don't know about Vishera, not yet anyway. We don't know what the improvements over Bulldozer will yield as a whole, only what a couple of benchmarks showed in a brief Toms comparison between Trinity and Zambezi. There are plenty of scenarios to consider.Beenthere - Thursday, September 27, 2012 - link
Yes some of us do know the results... Comparing Trintiy to Vishera is incorrect. Vishera is to be compared to Zambezi.AMD has hit their projected 10-15% gains for Vishera compared to Zambezi. Some people already know the results but the NDA doesn't expire for a few weeks so they can't print them yet. Most folks will be happy with Vishera except the haters.
silverblue - Thursday, September 27, 2012 - link
I'd find it hard to believe you were personally under NDA (please prove me wrong). I also believe the gains were per clock, which should theoretically, given the assumption you stated, result in a slightly larger performance gap between the 8150 and the 8350 as the latter has a higher base clock and is more likely to hit max turbo speed.Like I said though, two benchmarks in the public domain aren't gospel, regardless of whether we're comparing Vishera OR Trinity to Zambezi. Remember that L3 cache doesn't always help, but when it does, the gains can be significant, meaning the A10-5800K could occasionally be outperformed by a similarly clocked 41x0 CPU, but the flip side is that it could occasionally perform on par with a similarly clocked 43x0 CPU.
Death666Angel - Thursday, September 27, 2012 - link
"AMD was a little late in getting to the CPU - GPU party. Their first endeavour, the Llano APU"Aren't Zacate and Ontario APUs? They were released in 01/2011, half a year before Llano. Or aren't you counting low power APUs? :)
Thanks for the article!
Death666Angel - Thursday, September 27, 2012 - link
Mea culpa, didn't read the comments before posting my own. :) Disregard.EnzoFX - Thursday, September 27, 2012 - link
What's the big deal with 4k at THIS moment? There are no 4k tv's out are there? By the time they're out, or by the time they're actually affordable by a decent amount of consumers, we would have several generations of new apu's.Denithor - Tuesday, October 2, 2012 - link
http://www.lg.com/us/tvs/lg-84LM9600-led-tvAllio - Thursday, September 27, 2012 - link
These HTPC-perspective articles are consistently some of the most useful and interesting content that AT puts up. As far as I can tell, there really aren't any other tech sites that delve this deep into this kind of functionality - most reviews settle for playing a 1080p Bluray and posting a screenshot of the CPU usage in task manager. While it may only be a relatively small audience for who this stuff is relevant, we are a very interested audience, and I personally appreciate every detail and statistic included. Thanks Ganesh!wharris1 - Thursday, September 27, 2012 - link
At this point, it seems 4k is more marketing hype. I'll link this article: http://reviews.cnet.com/8301-33199_7-57491766-221/...For the typical anandtech readers (probably much more technically gifted than me) I also recall reading a similar article/post on avsforums explaining that for any display size <~100 inches, the 4k standard is hard to justify. Also, while I know that future proofing is sound, there is very little content or ability to play back said content at that resolution. As a previous poster mentioned, by the time 4k becomes a standard, the current platforms will seem antiquated. Anyway, Anandtech is the best tech site around by far; read it every morning.