Over the last few months or AMD has been slowly dropping hints about their upcoming APU and CPU revisions. In the interim there have been some updates and a pretty nice surprise in the budget-oriented area. While many have focused upon the upcoming Zen microarchitecture, exactly where it would fit into the 2016 (and beyond) product stack wasn’t exactly evident. In another one of the fits and spurts of information, AMD has begun clarifying at least one part of their future indicatives: the Bristol Ridge platform.
If you remember some of the previous articles I wrote about this topic, AMD’s future APUs and general processing cores will be consolidated into a single socket type. Dubbed AM4, this socket will be the foundation upon which many (but not all) products will be built upon. As with previous generations there will be additional embedded solutions that won’t come with a socketed design but discussing those will have to wait for a little while since they’re currently residing “beyond the horizon”.
That AM4 socket will actually be utilized in two separate platforms: Bristol Ridge and Summit Ridge, the former of which will be used for upcoming APUs. Summit Ridge meanwhile will be very much targeted towards standard next generation CPUs but those are still some time away. Meanwhile, the focus here is its notebook iteration which is meant to replace and supersede existing Carrizo APUs.
With this background story now common knowledge we can get to the star of this particular article: the first entries under the Bristol Ridge umbrella. These “7th generation APUs” will feature an updated version of the Excavator architecture used within the Carrizo architecture with significantly higher single thread x86 performance and a DDR4 memory controller. According to AMD there will also be an optimized GPU core and more design changes to further refine performance per watt but what those entail isn't being discussed right now.
Unfortunately, AMD won’t silence the rumor mill since there’s still a lot of information being left behind closed doors. There has been no mention of which manufacturing process these Carrizo derivatives use, nor is there any indication of the GPU architecture or what the actual lineup will look like or frequencies. If I had to make some educated guesses at the first two questions, those answers would be: 28nm and GCN 1.2 or the artist formerly known as Volcanic Islands which graces the Tonga and Fiji GPUs.
If this sounds familiar that’s because many of these features have already been rolled into AMD’s Merlin Falcon embedded R-Series APUs which are based directly off of the Carrizo core design. From revised TDP metrics, to DDR4 support to GCN 1.2-based GPUs; its all been there for a while, though not available on mass market products. The 7th generation APUs seem to be socketed and slightly faster versions of the RX-421BD and its ilk, though there may be some fine-grain differences.
All in all, this looks like a pipe-cleaning launch in preparation for a broad move beyond Excavator and into the evolutionary (and hopefully revolutionary) Zen generation. While the desktop market has seen a simple lone Excavator-based Anthon SKU, we also can’t forget that Carrizo has been available to notebook manufacturers for the better part of a year now. That means it is high time for a FX-series revision targeting the mobile segment in an effort to keep AMD’s offerings in contention against Intel’s Skylake alternatives.
Within the slide deck I was given last night, AMD added a few interesting benchmark results but, as is usual for any marketing-focused press materials the really interesting information lies below the surface. After digging a bit deeper into the fine print, the actual test systems being used are virtually identical minus one critical bit: the newest APU was always run using higher memory speeds.
In the charts below I’ve compiled testing benchmark results with their true representative scores as displayed by AMD’s official testing procedures. It’s important to remember that these results are AMD’s own internal numbers and have not been validated by Hardware Canucks, nor do we know the actual clock speeds of these new parts so comparative analysis will have to wait for another day.
The first benchmark is one in which AMD’s powerful onboard GPUs have always shone: 3DMark. Here the 7th generation APU is really able to stretch its legs with huge performance benefits over Intel’s i7-6500U and the previous 6th generation Carrizo-based FX-8800P. One interesting thing to note is that both APUs are running at their lower 15W TDP setting rather than their configurable 35W limit.
I would love to say that all of AMD’s newfound performance is due to a revised architecture but as we’ve seen in the past, the benefit of running 1866MHz memory versus the 1600MHz of the other two chips just can’t be overlooked. Now there’s certainly some optimizations going on here but APUs absolutely love additional bandwidth when it comes to graphics processing.
PCMark sees the Bristol Ridge APU coming out with moderate increases over a first generation Carrizo APU despite its faster memory configuration. Also noteworthy is the fact that AMD chose to run the Accelerated version of this benchmark which plays to their APUs’ strengths since it engages the GPU compute engine.
I also want to point out how well it looks like AMD is closing the gap between their current generation Excavator architecture and Intel’s Skylake, at least on the notebook platform. Being within about 5% on a 14nm processor that has an identical TDP rating is no small thing and if these results truly translate into real-world ones, AMD’s customers have a lot to be excited about.
Whereas graphics benchmarks have always been a strength of AMD’s APUs, their architectures have struggled with single thread workloads ever since Bulldozer’s modularized design was first introduced. With that in mind, the engineers have been hard at work optimizing Piledriver, Steamroller and now Excavator for increased throughput in simple x86 processing scenarios. The 7th generation APUs obviously have some very serious advances built into their designs since benchmarks point towards a nearly 20% uplift.
Now some of you may once again point towards the 2400MHz(!!) memory speed but as has been proven time and again, Cinebench doesn’t really benefit all that much from higher memory bandwidth. As such, this could be down to anything from improved TDP management so Turbo algorithms consistently hit maximum targets or just a matter of better instruction set handling. There is however a small caveat here: all of the solutions were pushed to their 35W maximum so it is quite likely they’d still be overcome by the 15W i7-6500U from previous benchmarks which levels out around the 127 mark.
Moving onto the multi threaded testing its obvious that the Excavator architecture brings with it some borderline incredible performance benefits when compared to Steamroller. Meanwhile, the latest 7th generation APUs take things to another level by improving upon Carrizo by another 10%. This is the kind of benchmark result desktop users have been yearning for so it should be interesting to see what happens if or when AMD ends up bringing this new design over to their other platforms.
While AMD’s slow and steady release of information about their upcoming platforms can be considered a patchwork of raw information interspersed by no small amount of understandable rumor mongering, it is good to see some forwards progress. Desktop users may have to wait a while longer to see anything concrete materialize in their end of the computing spectrum but there are some very encouraging –though very early- signs of hope based on what is being shown in the notebook space.
Speaking of notebooks, that’s where AM4 / Bristol Ridge will obviously be targeted with the refreshed and reinvigorated Carrizo design covered in this article. As a matter of fact AMD is supposedly ahead of time for production on these 28nm chips, with Computex serving as an official launch platform for numerous designs based on Bristol Ridge. Stay tuned for that but until then, AMD has a “wait, there’s one more thing!” moment for us: The HP Envy X360.
While AMD wasn’t able to give an availability date, they did mention the X360 would be the first notebook to house their 7th generation APUs. It actually looks like a pretty decent convertible notebook with a 15.6” IPS screen, 10 hours of battery life, USB Type-C, a weight of just 4.8lbs and a thickness of 18.8mm. Alongside dual and quad core APU options there’s also the possibility of a 256GB SSD (or 2TB HDD) and an audio solution by Bang & Olufsen. This truly does look like a flagship showing for AMD’s updated architecture.
As with all of these previews, the success or failure of AMD’s newest initiatives will ultimately lie in how the market reacts to their offerings. It looks like steps are being taken in the right direction –the odd distribution of key information notwithstanding- but only time will tell how the 7th generation APUs perform against the competition and how long they’ll be in place before Zen is able to take up the torch. From what I have seen in these first benchmark results from AMD's own labs, there could be a lot to look forward to.
If you remember some of the previous articles I wrote about this topic, AMD’s future APUs and general processing cores will be consolidated into a single socket type. Dubbed AM4, this socket will be the foundation upon which many (but not all) products will be built upon. As with previous generations there will be additional embedded solutions that won’t come with a socketed design but discussing those will have to wait for a little while since they’re currently residing “beyond the horizon”.
That AM4 socket will actually be utilized in two separate platforms: Bristol Ridge and Summit Ridge, the former of which will be used for upcoming APUs. Summit Ridge meanwhile will be very much targeted towards standard next generation CPUs but those are still some time away. Meanwhile, the focus here is its notebook iteration which is meant to replace and supersede existing Carrizo APUs.
With this background story now common knowledge we can get to the star of this particular article: the first entries under the Bristol Ridge umbrella. These “7th generation APUs” will feature an updated version of the Excavator architecture used within the Carrizo architecture with significantly higher single thread x86 performance and a DDR4 memory controller. According to AMD there will also be an optimized GPU core and more design changes to further refine performance per watt but what those entail isn't being discussed right now.
Unfortunately, AMD won’t silence the rumor mill since there’s still a lot of information being left behind closed doors. There has been no mention of which manufacturing process these Carrizo derivatives use, nor is there any indication of the GPU architecture or what the actual lineup will look like or frequencies. If I had to make some educated guesses at the first two questions, those answers would be: 28nm and GCN 1.2 or the artist formerly known as Volcanic Islands which graces the Tonga and Fiji GPUs.
If this sounds familiar that’s because many of these features have already been rolled into AMD’s Merlin Falcon embedded R-Series APUs which are based directly off of the Carrizo core design. From revised TDP metrics, to DDR4 support to GCN 1.2-based GPUs; its all been there for a while, though not available on mass market products. The 7th generation APUs seem to be socketed and slightly faster versions of the RX-421BD and its ilk, though there may be some fine-grain differences.
All in all, this looks like a pipe-cleaning launch in preparation for a broad move beyond Excavator and into the evolutionary (and hopefully revolutionary) Zen generation. While the desktop market has seen a simple lone Excavator-based Anthon SKU, we also can’t forget that Carrizo has been available to notebook manufacturers for the better part of a year now. That means it is high time for a FX-series revision targeting the mobile segment in an effort to keep AMD’s offerings in contention against Intel’s Skylake alternatives.
Performance Metrics Explored
Within the slide deck I was given last night, AMD added a few interesting benchmark results but, as is usual for any marketing-focused press materials the really interesting information lies below the surface. After digging a bit deeper into the fine print, the actual test systems being used are virtually identical minus one critical bit: the newest APU was always run using higher memory speeds.
In the charts below I’ve compiled testing benchmark results with their true representative scores as displayed by AMD’s official testing procedures. It’s important to remember that these results are AMD’s own internal numbers and have not been validated by Hardware Canucks, nor do we know the actual clock speeds of these new parts so comparative analysis will have to wait for another day.
The first benchmark is one in which AMD’s powerful onboard GPUs have always shone: 3DMark. Here the 7th generation APU is really able to stretch its legs with huge performance benefits over Intel’s i7-6500U and the previous 6th generation Carrizo-based FX-8800P. One interesting thing to note is that both APUs are running at their lower 15W TDP setting rather than their configurable 35W limit.
I would love to say that all of AMD’s newfound performance is due to a revised architecture but as we’ve seen in the past, the benefit of running 1866MHz memory versus the 1600MHz of the other two chips just can’t be overlooked. Now there’s certainly some optimizations going on here but APUs absolutely love additional bandwidth when it comes to graphics processing.
PCMark sees the Bristol Ridge APU coming out with moderate increases over a first generation Carrizo APU despite its faster memory configuration. Also noteworthy is the fact that AMD chose to run the Accelerated version of this benchmark which plays to their APUs’ strengths since it engages the GPU compute engine.
I also want to point out how well it looks like AMD is closing the gap between their current generation Excavator architecture and Intel’s Skylake, at least on the notebook platform. Being within about 5% on a 14nm processor that has an identical TDP rating is no small thing and if these results truly translate into real-world ones, AMD’s customers have a lot to be excited about.
Whereas graphics benchmarks have always been a strength of AMD’s APUs, their architectures have struggled with single thread workloads ever since Bulldozer’s modularized design was first introduced. With that in mind, the engineers have been hard at work optimizing Piledriver, Steamroller and now Excavator for increased throughput in simple x86 processing scenarios. The 7th generation APUs obviously have some very serious advances built into their designs since benchmarks point towards a nearly 20% uplift.
Now some of you may once again point towards the 2400MHz(!!) memory speed but as has been proven time and again, Cinebench doesn’t really benefit all that much from higher memory bandwidth. As such, this could be down to anything from improved TDP management so Turbo algorithms consistently hit maximum targets or just a matter of better instruction set handling. There is however a small caveat here: all of the solutions were pushed to their 35W maximum so it is quite likely they’d still be overcome by the 15W i7-6500U from previous benchmarks which levels out around the 127 mark.
Moving onto the multi threaded testing its obvious that the Excavator architecture brings with it some borderline incredible performance benefits when compared to Steamroller. Meanwhile, the latest 7th generation APUs take things to another level by improving upon Carrizo by another 10%. This is the kind of benchmark result desktop users have been yearning for so it should be interesting to see what happens if or when AMD ends up bringing this new design over to their other platforms.
Some Closing Thoughts
While AMD’s slow and steady release of information about their upcoming platforms can be considered a patchwork of raw information interspersed by no small amount of understandable rumor mongering, it is good to see some forwards progress. Desktop users may have to wait a while longer to see anything concrete materialize in their end of the computing spectrum but there are some very encouraging –though very early- signs of hope based on what is being shown in the notebook space.
Speaking of notebooks, that’s where AM4 / Bristol Ridge will obviously be targeted with the refreshed and reinvigorated Carrizo design covered in this article. As a matter of fact AMD is supposedly ahead of time for production on these 28nm chips, with Computex serving as an official launch platform for numerous designs based on Bristol Ridge. Stay tuned for that but until then, AMD has a “wait, there’s one more thing!” moment for us: The HP Envy X360.
While AMD wasn’t able to give an availability date, they did mention the X360 would be the first notebook to house their 7th generation APUs. It actually looks like a pretty decent convertible notebook with a 15.6” IPS screen, 10 hours of battery life, USB Type-C, a weight of just 4.8lbs and a thickness of 18.8mm. Alongside dual and quad core APU options there’s also the possibility of a 256GB SSD (or 2TB HDD) and an audio solution by Bang & Olufsen. This truly does look like a flagship showing for AMD’s updated architecture.
As with all of these previews, the success or failure of AMD’s newest initiatives will ultimately lie in how the market reacts to their offerings. It looks like steps are being taken in the right direction –the odd distribution of key information notwithstanding- but only time will tell how the 7th generation APUs perform against the competition and how long they’ll be in place before Zen is able to take up the torch. From what I have seen in these first benchmark results from AMD's own labs, there could be a lot to look forward to.
Last edited:
