Subject: General Tech, Graphics Cards, Processors | August 22, 2013 - 01:39 PM | Ryan Shrout
Tagged: sony, ps4, playstation 4, Kabini, hUMA, amd
UPDATE: I have added new info at the bottom of this post with more commentary from AMD (kind of).
You might have seen some reports in the last couple of days claiming that the upcoming Sony PlayStation 4 (PS4) will have a big advantage over the Xbox One thanks to its unique ability to support AMD's hUMA memory architecture. hUMA, heterogeneous unified memory architecture, is an exciting new memory technology that AMD has built into upcoming APUs.
Josh published a story on hUMA that sums it as so:
The idea behind hUMA is quite simple; the CPU and GPU share memory resources, they are able to use pointers to access data that has been processed by either one or the other, and the GPU can take page faults and not rely only on page locked memory. Memory in this case is bi-directionally coherent, so coherency issues with data in caches which are later written to main memory will not cause excessive waits for either the CPU or GPU to utilize data that has been changed in cache, but not yet written to main memory.
There's just one problem with these various reports (VR-Zone, ExtremeTech): they're incorrect. After sending some emails to our representatives at AMD I was told that "Kabini doesn't support hUMA" which is the APU that both the PS4 and Xbox One processors are based on. AMD further clarified with us:
Our spokesperson made inaccurate statements about our semi-custom APU architectures and does not speak for Microsoft, Sony or the AMD semi-custom business unit responsible for co-developing the next generation console APUs.
So while the PS4 will still be a faster system thanks to its higher SIMD processor (GPU core) count, there is no support for a true heterogeneous unified memory architecture in either upcoming console platform.
NOTE: I have had several people point out that it's possible Sony and Microsoft worked on their own custom memory architectures that will perform similar functionally to hUMA. That is entirely possible but means that official hUMA support isn't on the SoCs.
UPDATE: AMD contacted me again to make another comment. Essentially, they said that the correction statement to the original statement claiming hUMA was part PS4 was "inaccurrate" but that this correction does NOT mean the opposite claim is true. Even when pressed for a more specific and debate-ending comment, AMD wouldn't give us any more information.
So does the PS4 have support for some type of heterogeneous unified memory? Maybe. And the Xbox One? Maybe. At this point, I'd stop listening to anything AMD has to say on the subject as they are likely to recant it shortly thereafter. Many readers have emailed me with their thoughts and I personally feel that its more likely the original statement from AMD (that the PS4 will have the edge with a hUMA design) will turn out to be the truth in the long run...
Subject: General Tech, Graphics Cards, Processors | August 16, 2013 - 04:00 PM | Scott Michaud
Tagged: nvidia, Intel, APU, amd
Despite a slight decline in PC sales compared to last quarter, graphics processors are on the rise. Jon Peddie Research attributes the heightened interest in graphics, with a decline in systems, to a trend towards multiple GPUs in a system. Crossfire and SLI, according to the report, are not driving this drift but they are relevant. More importantly, consumers are adding discrete graphics to systems with integrated solutions.
AMD has experienced an increase in shipments of 47% for laptop APUs. Desktop heterogeneous processors declined but, in all, shipments increased 11%. Intel, likewise, saw an increase albeit just 6%. NVIDIA declined 8%. AMD now enjoys a 5.8% lead in total market share over NVIDIA.
Many PCs have access to multiple graphics processors simultaneously. With an increase of available GPUs, software developers might take the plunge into fully supporting heterogeneous architectures. You could imagine a game which offloads physics or AI pathfinding to secondary graphics. Sure, the increased heat would slightly limit the turbo-performance of the CPU, but the increased parallel performance should overtake that decreased serial performance for a sensible developer.
JPR claims an average of nearly 1.4 GPUs available per system.
The increased laptop heterogeneous processors is a major win for AMD. Still, I wonder how much Never Settle played in to users dropping discrete graphics into machines which would otherwise have integrated (chipset or processor) graphics. The discrete graphics market has declined and yet somehow AMD got a boost from double-attach or replaced graphics.
The report only discusses consumer x86 tablets, desktops, laptops, and some hybrid between the previous three categories. Other processor architectures or x86 servers are not covered.
Subject: General Tech, Processors | August 11, 2013 - 04:26 PM | Scott Michaud
Tagged: Ivy Bridge-E, Intel
Ivy Bridge was well known, just not in a good way, for its overclocking ability. We noted how sharply temperatures rose when frequencies were increased above factory recommendations in our i7 3770K review. Performance scaled well but, even with a decent aftermarket cooler, only did so while close to the boiling point of water. As Ryan described it,
To be fair, the 1.3v setting for this processor is on the upper limit of what you should be using according to many reports. The 22nm process is great for low power consumption but apparently not great for overclocking - higher voltages result in much higher temperatures than what we would have seen on Sandy Bridge.
...and a 24% boost in TrueCrypt. Pretty impressive results actually. But things are getting HOT under our Corsair H80 as it was unable to keep the CPU from breaching the 80C mark.
According to a roadmap received by VR-Zone, the pinnacle of Ivy Bridge-E until at least Q2 2014 will be the Core i7-4960X which, last month, failed to excite enthusiasts when benchmarks leaked. The story kept true from mainstream: it remains in Sandy Bridge-E's ballpark but requires less power. I am sure that Amazon Web Services will be thrilled...
We have wondered if Intel intends to punt this launch, fulfill commitments to Socket 2011 and nothing more, in preparation of Haswell-E. We may never see the i7-4960X overthrown, Xeon notwithstanding, until after the socket is retired.
But, now, we get to the hopeful news.
Unlike the prior generation, Sandy Bridge-E, the i7-4960X will not be a crippled Xeon architecture with disabled cores. While still a 6-core part, it will be so natively. Previously, the 6-core Sandy Bridge-E was an 8-core product with two disabled. This is an advantage because, assuming the locked cores could never be restored, their absence should allow greater overclocking headroom. Factor in the quad-channel DDR3-1866, which itself should have decent overclock potential, and users might have more room to be enthusiastic enthusiasts.
Overclocking capacity was the biggest unknown from last month's leaks. It is now looking a little more hopeful, at least for those with Sandy Bridge-E and an intent to replace their CPU before their motherboard.
And the pricing...?
According to the above table, originally from VR-Zone, the top two Ivy Bridge-E SKUs are expected to come in cheaper by $50-$70 than the Sandy Bridge-E models they retire. The quad-core i7-4820K is the exception, being priced within $5 of its ancestor.
Ivy Bridge-E is expected to launch in just a couple of months.
Subject: Editorial, General Tech, Processors, Mobile | August 3, 2013 - 07:21 PM | Scott Michaud
Tagged: qualcomm, Intel, mediatek, arm
MediaTek, do you even lift?
According to a Taiwan Media Roundtable transcript, discovered by IT World, Qualcomm has no interest, at least at the moment, in developing an octo-core processor. MediaTek, their competitor, recently unveiled an eight core ARM System on a Chip (SoC) which can be fully utilized. Most other mobile SoCs with eight cores function as a fast quad-core and a slower, but more efficient, quad-core processor with the most appropriate chosen for the task.
Anand Chandrasekher of Qualcomm believes it is desperation.
So, I go back to what I said: it's not about cores. When you can't engineer a product that meets the consumers' expectations, maybe that’s when you resort to simply throwing cores together. That is the equivalent of throwing spaghetti against the wall and seeing what sticks. That's a dumb way to do it and I think our engineers aren't dumb.
The moderator, clearly amused by the reaction, requested a firm clarification that Qualcomm will not launch an octo-core product. A firm, but not clear, response was given, "We don't do dumb things". Of course they would not commit to swearing off eight cores for all eternity, at some point they may find core count to be their bottleneck, but that is not the case for the moment. They will also not discuss whether bumping the clock rate is the best option or whether they should focus on graphics performance. He is just assured that they are focused on the best experience for whatever scenario each product is designed to solve.
And he is assured that Intel, his former employer, still cannot catch them. As we have discussed in the past: Intel is a company that will spend tens of billions of dollars, year over year, to out-research you if they genuinely want to play in your market. Even with his experience at Intel, he continues to take them lightly.
We don't see any impact from any of Intel's claims on current or future products. I think the results from empirical testers on our products that are currently shipping in the marketplace is very clear, and across a range of reviewers from Anandtech to Engadget, Qualcomm Snapdragon devices are winning both on experience as well as battery life. What our competitors are claiming are empty promises and is not having an impact on us.
Qualcomm has a definite lead, at the moment, and may very well keep ahead through Bay Trail. AMD, too, kept a lead throughout the entire Athlon 64 generation and believed they could beat anything Intel could develop. They were complacent, much as Qualcomm sounds currently, and when Intel caught up AMD could not float above the sheer volume of money trying to drown them.
Then again, even if you are complacent, you may still be the best. Maybe Intel will never get a Conroe moment against ARM.
Subject: Processors | July 31, 2013 - 06:47 AM | Tim Verry
Tagged: lga 2011, Ivy Bridge-E, Intel
Intel’s Ivy Bridge-E processors are on their way to enthusiasts and should be available as soon as September 4th according to TechPowerUp. The new HEDT parts are compatible with LGA 2011 motherboards and the CPUs bring performance and power efficiency increases to the enthusiast platform.
The Ivy Bridge-E parts start at $310 and will be price competitive with the existing Sandy Bridge-E and high end Haswell parts. The Core i7-4820K is the lowest-end Ivy Bridge-E processor. This quad core part has base and turbo clockspeeds of 3.7 GHz and 3.9 GHz respectively, 10MB of L3 cache, 48 PCI-E 3.0 lanes, and a quad channel memory controller. It will cost $310.
Beyond that, the Core i7-4930K is the lowest-end six core IVB-E processor. It has six cores clocked at 3.4 GHz base and 3.9 GHz turbo, 12MB L3 cache, 48 PCI-E 3.0 lanes, and a quad channel memory controller. This part will cost $555.
Finally, the top end Core i7-4960X Extreme Edition is a six core CPU clocked at 3.6 GHz base and 4.0 Ghz turbo with 15MB of L3 cache, 48 PCI-E 3.0 lanes, and a quad channel memory controller. This Ivy Bridge-E HEDT CPU will cost $990.
Compared to the existing Sandy Bridge-E chips, IVB-E is actualy coming out of the gate with lower initial MSRPs. Further, the i7-4820K is actually about $30 cheaper than the Core i7-4770K “Haswell” CPU. The six core i7-4930K may be more enticing to those comparing the enthusiast LGA 2011 platform and the LGA 1150 platform, however. The six core Ivy Bride-E part is about $215 more expensive than the $340 four core Haswell i7-4770K which may be a small enough gap that enthusiasts are willing to make the jump for the extra two cores. Granted, Haswell is a bit faster in some respects than IVB-E (according to leaked benchmarks), but the extra two cores gives it a healthy multi-threaded performance advantage.
Read more about Intel's upcoming Ivy Bridge-E processors.
Subject: Processors | July 31, 2013 - 03:59 AM | Tim Verry
Tagged: Kaveri, fm2, carrizo, APU, amd radeon, amd
Rumors recently surfaced via VR-Zone china that AMD’s Kaveri APU successor will be code-named Carrizo, and it will be compatible with the upcoming FM2+ socket and AMD A88X chipset that Kaveri will use.
AMD’s Carrizo APUs will reportedly be available in TDPs up to 65W and will feature Excavator CPU cores along with a next generation Radeon GPU. Much like Kaveri, Carrizo will be fully HSA compliant. The chips will also include support for DDR4.
Carrizo will allegedly begin sampling in August 2014 with mass production starting around December. That means Carrizo will be available for purchase within the first half of 2015.
FM2+ boards like the ASUS A55BM-A/USB3 are rumored to support AMD's Carrizo APUs (the successor to Kaveri).
The rumors also suggest that Carrizo will be joined by a low power “Beema” System on a Chip (SoC) and a BGA-based Nolan APU for embedded systems. Details on these complementary chips are scarce, however. Perhaps most telling is last bit of the article that suggests that AMD will not be releasing a AM3+ Vishera CPU-only processor successor. It seems AMD is going all in as an APU company after all.
I have been looking forward to the launh of AMD's Kaveri since AFDS 2012, and Carrizo appears to be a refinement of that chip. It should be more power efficient and faster thanks to architecture tweaks and process shrinks. I think that AMDs architecture and HSA approach has potential, and I'm excited to see what these upcoming chips can do with regards to performance.
Subject: Motherboards, Processors | July 30, 2013 - 05:00 AM | Tim Verry
Tagged: P9X79, Ivy Bridge-E, IVB-E, bios, ASUS ROG Rampage IV GENE X79, asus
ASUS has released BIOS updates for all of its LGA 2011 X79 motherboards that add support for Intel's upcoming Ivy Bridge-E processors. The update enables enthusiasts to upgrade from their existing Sandy Bridge-E CPU to an Ivy Bridge-E model which adds marginal improvements in performance and power efficiency. Supported processors include the Core i7-4960X Extreme Edition, Core i7-4930K, and the Core i7-4820K. According to benchmarks by Tom's Hardware, the top-end Ivy Bridge-E i7-4960X exhibits up to 30% improvements in performance per watt along with being slightly faster in multi-threaded performance than SB-E. Of course, single threaded performance was shown to be similar to that of Sandy Bridge-E but slower than Haswell.
All of ASUS' X79 boards (including the ROG Rampage, TUF, and P9X79 series) will be getting a BIOS update which will be made available for download on the company's support website or via the individual motherboard product pages. The following chart indicates the motherboard and associated BIOS version number that adds support for IVB-E.
|Motherboard||BIOS Version Supporting IVB-E|
|ROG Rampage IV Extreme||4206|
|ROG Rampage IV Formula||4004|
|ROG Rampage IV Gene||4206|
|TUF Sabertooth X79||4104|
To grab the update, head over to the ASUS support site and type in your motherboard model name to get the appropriate BIOS file.
Will you be upgrading your LGA 2011 rig to Ivy Bridge-E?
Subject: Processors | July 28, 2013 - 01:08 PM | Tim Verry
Tagged: Richland, overclocking, LN2, APU, amd, a10-6800k
A Finnish overclocker known as “The Stilt” recently pushed an AMD Richland APU to 8.2GHz using liquid nitrogen. In doing so, The Stilt broke the world record for APU overclocking, besting his previous overclock attempt.
Specifically, the chip was a retail version of the AMD A10-6800K “Richland” APU. It was overclocked to 8203.01 MHz with a 130.21 MHz base clock and 63x multiplier. Even more impressive is that The Stilt managed the overclock with less voltage -- 1.968 volts -- than his earlier (and lower) overclock. For comparison, the earlier overclock brought the A10-6800K to 8000.48 MHz using 2.008 volts.
The system used to overclock the APU included an ASUS F2A85-V Pro motherboard, 8GB of AMD DDR3 Performance memory, and a Radeon HD 7750 graphics card. The overclocker used liquid nitrogen to cool the APU while the GPU was left at stock settings and with its default air cooler. The RAM was overclocked to 2083.6 MHz with 10-11-10-27 timings.
In all, it is an impressive overclock considering all four CPU cores were left enabled! More details along with validation of the overclock can be found over at HWBot.
Also read: AMD A10-6800K and A10-6700 Review: Richland Finally Lands @ PC Perspective
Subject: General Tech, Processors, Mobile | July 23, 2013 - 06:58 PM | Scott Michaud
Tagged: SDP, haswell
Intel has just lowered their lowered thermal expectations for Haswell if, of course, you use SDP as your metric. Scenario Design Point (SDP), as opposed to Thermal Design Power (TDP), describes how much heat dissipation is required for the product at some, usually underclocked, performance target. SDP does not need to affect burst performance, however, as the chip can still up-clock given some extra headroom.
While we don't know OEM partners, Intel could be green, with HP Envy?
It describes long-term cooling requirements, not instantaneous power draws.
In terms of SDP, Intel expected to ship 6W products based on their 4th generation core architecture. Today, Intel announced a limited stock will dip below that target, capable of just 4.5W in waste heat. OEMs who purchase from this limited binning will be able to include Haswell in even thinner active or passively cooled designs.
Intel has not described exact specifications, partners, or shipping dates.
Subject: General Tech, Graphics Cards, Processors, Mobile | July 23, 2013 - 04:01 AM | Scott Michaud
Tagged: Samsung, mali, exynos
Exynos, the line of System on a Chip (SoC) products from Samsung, were notably absent of ARM Mali GPUs. This, apparently, struck concern over how viable Mali will continue to be and whether ARM will continue to lose designs to competitors such as Imagination Technologies.
Then Samsung announced, Monday evening for us North Americans, the upcoming Exynos 5 Octa Processor will embed six ARM Mali-T628 GPU cores. The T628 GPU cores are capable of OpenCL 1.1 and OpenGL ES 3.0 standards which should allow applications to offload heavy batches of tasks, such as computational photography processing, with high efficiency and performance.
The Exynos 5 Octa contains four ARM Cortex-A15 cores at 1.8GHz, supported by four additional Cortex-A7 cores clocked at 1.3GHz. These processors are currently being sampled and should be produced in August.