Processors - Page 117

AMD's new APUs are pretty awesome. They combine great graphics with pretty good compute performance into a single chip. They are ideal for laptops, desktops, and embedded solutions. Today's award announcement, which brings them to their second consecutive win, is given by Taipei Computer Association after being judged by a panel of esteemed government officials, academics, research analysts, editors-in-chief of key media outlets and industry experts.

"We are honored to receive this prestigious award from the Taipei Computer Association for the second consecutive year," said Chris Cloran, corporate vice president, AMD Client Products. "The overwhelming market reception and industry accolades for the AMD A-Series APUs demonstrate that AMD is uniquely meeting the needs of our customers and a new generation of PC buyers."

According to the press release:

The AMD FX series of processors is known for overclocking pretty well. It currently holds the world record for fastest clock speed ever achieved and looks to be breaking that once again. A new CPU-z validation screen shot has shown up online. It details the feat even though HWBot rejected the the posting.

The rejection was most likely due to the fact that the overclocker didn't provide much in the way of detail about the cooling system used. When I say much, I mean any. What we do know is that the chip was clocked at a massive 8805.6MHz and was achieved by overclocker Ksin. Speed like that comes only when an overclocker uses extreme cooling such as liquid nitrogen or liquid helium.

To achieve the feat, Ksin used a base clock of 303.29 MHz and a multiplier of 29.0. To reach these clock speeds, an incredible 1.86 volts were pumped through the core. As such, it is recommended that you don't try this at home as that will kill your chip very quickly. Other components of the system include an ASUS Crosshair V Formula motherboard and 4 GB (2x 2 GB) of ADATA memory. The memory was clocked at 1417 MHz.

"Industry sources" are reporting that the next-generation AMD architecture code-named "Piledriver" will commence production in Q3 2012. Bulldozer, Piledriver's predecessor, was only released in October of 2011, so this could match Intel's one year release cycle. Trinity, an APU based upon Piledriver was released just last week.

The upcoming CPU will continue to be based upon the AM3+ platform. This simple fact should keep enthusiasts happy and will allow for current Bulldozer users to upgrade for only the cost of a CPU. AMD has always had a history of retaining sockets and providing upgrade paths whereas Intel often kills off sockets with each new processor.

Piledriver will be very similar to Bulldozer in that it will be made on a 32nm process and feature the same compute module design. Piledriver will also retain the same dual-channel 1866MHz DDR3 memory controllers. Piledriver will continue to feature AMD's Turbo Core at version 3.0. The new CPU will deliver more performance, thanks to an improved IPC, at lower power usage, thanks to AMD's new energy-recycling technology called resonant clock mesh.

If we remember late last year, AMD had talked about the transistor count in their then-released Bulldozer chip, which was meant to house 2 billion transistors, but actually only had 1.2 billion. How a mistake like that can happen is anyone's guess, but it looks like Team Red are up to their old tricks again, it seems.

Bright Side of News has an interesting piece, where they've noticed that AMD have again given the public an erroneous transistor count. AMD's Llano was meant to have 1.45billion transistors, as the company explained at the Hot Chips 2011 conference, as well as most reviews citing that it sported 1.45 billion transistors.

But, if we look at some of the Trinity reviews, such as AnandTech's, where they state it has 1.178 billion transistors, Tom's Hardware and Tech Report state it has 1.45 billion, but Trinity actually sports 1.30 billion transistors. The die size hasn't changed, and measures in at the same 228mm² since Llano emerged in 2011. Most sites can't really be faulted here, as they've just copy/pasted the specs directly from... AMD.

Researchers from Rice University in Houston, Singapore's Nanyang Technological University (NTU), Switzerland's Center for Electronics and Microtechnology (CSEM) and the University of California, Berkeley have unveiled an "inexact" chip at the ACM International Conference on Computing Frontiers in Cagliari, Italy.

The chip is allowed to make mistakes in order to gain speed and energy usage advances. This new "inexact" processor is up to 15 times more efficient than current processors. It can be used in certain applications where 100% accuracy isn't mission critical. Examples of this would be video or picture processing.

"It is exciting to see this technology in a working chip that we can measure and validate for the first time," said project leader Krishna Palem, who also serves as director of the Rice-NTU Institute for Sustainable and Applied Infodynamics (ISAID). "Our work since 2003 showed that significant gains were possible, and I am delighted that these working chips have met and even exceeded our expectations."

Intel really wants to cosy up to Apple, with CEO Paul Otellini having made it very clear to investors and reporters that the chip-making company will continue to angle more business from Apple in mobile devices, with his mission to "ensure our silicon is so compelling [that] they can't ignore us".

Intel already have a strong relationship with Apple as they're the ones baking processors in all of Apple's desktop and notebook products, but Intel want to replace the ARM-based processors that you'll find in Apple's mobile lineup.

If we rewind a few years, we would have found stories of the iPod and iPhone architect Tony Fadell threatening to quite Apple in 2006 if the company had chosen to go with Intel for their forthcoming iPhone, where CEO Steve Jobs had actually favored. Intel's offerings at the time were not energy-efficient enough, but this was 2006 after all.

There has been a lot of debate back and forth about whether or not Ivy Bridge's "slight" heat problem is caused by Intel's choice to not use solder to attach the IHS to the core. Originally, people were blaming the higher density of the transistors, due to the 22nm process, for the heat issue. Then, someone decided to de-lid the chip and found TIM instead of solder.

First, let me recommend you not remove the IHS from your chip as it could damage your chip, not to mention it voids your warranty. Besides, someone else has already done it for you along with the testing needed to prove that the heat issue is due to the TIM used by Intel. Impress PC Watch wiped off Intel's TIM and replaced it with OCZ's Freeze Extreme and Coollaboratory's Liquid Pro to see the difference.

The results? As you can expect from the title of this news post, the TIM choice that Intel made did in fact yield higher temperatures. Load temperatures at stock settings results in an 8*C and 11*C drop in temperatures which is a pretty big change. Overclocking the chip to 4.6GHz resulted in even bigger differences: 15*C for the OCZ TIM and 20*C for the Coollaboratory TIM.

With the Trinity APU launch coming soon, leaks will be coming more frequently and with more credibility. The leak today is from AMD's website itself (the file has since been removed) in the form of a PDF file which had some characteristics such as model numbers and graphics performance of the upcoming APUs.

The part of interest is the graph of the upcoming APUs' graphics performance relative to Intel's Sandy Bridge CPUs. Why not Ivy Bridge? Well, the Trinity APUs shown in the graphic are embedded options and Intel hasn't launched any Ivy Bridge embedded options so this is a more apples-to-apples comparison.

What the graph shows is that the AMD R-464L has 206%, and the R-272F has 145% better performance than Intel i7-2710QE embedded chip in 3DMark 06 and Vantage v1.1.0 applications. However, that Intel chip is using the HD 3000 graphics which is noticeably slower than the HD 4000 graphics of Ivy Bridge.

On June 3, Intel plans to launch some cheaper dual-core Ivy Bridge-based CPUs, prices of these new chips will range between $225 and $346. We're also looking at two Core i7 options, the first being the Core i7-3520M which clocks in at 2.9GHz with a turbo frequency of 3.6GHz, 4MB of L3 cache, a 35W TDP and pricing of $346.

The second Core i7 offering, the Core i7-3667U retains the same L3 cache, but mixes up the clock speed to 2GHz, with a turbo frequency of 3.2GHz, but we have a big change in the TDP, just 17W, it also retains the same $346 pricing.

The three Core i3 offerings are split into 2 options, one of them with a 17W TDP, the remaining two with 35W TDPs. The first one, Core i5-3320M has a clock speed of 2.6GHz, turbo at 3.3GHz, 3MB of L3 cache and a price of $225, the second, Core i5-3360M has a frequency of 2.8GHz, turbo up to 3.5GHz, the same L3 cache amount, but a price of $266. The final Core i5-3427U has a clock speed of 1.8GHz, turbo at 2.8GHz, 3MB L3 cache, and a 17W TDP, its price is set at $225.

If you thought the dual-core processor in your smart device was fast, I bet you thought the quad-core processors we're seeing in smart devices like the Samsung GALAXY S III which sports the Exynos 4 Quad is impressive, well, not so much.

TSMC have just announced a 28nm ARM Cortex-A9 dual-core processor which can run at over 3GHz. These processors are from the high performance for mobile applications (HPM) process node, but will also be made to operate at lower speeds between 1.5GHz and 2.0GHz, for less demanding user markets. The high performance chips are to be baked into tablets, mobile products and networking applications.

TSMC has said that the resulting SoC designs will have the lowest PPA landmark (which is a ratio measure of power to area) available in the market. When compared to the 40nm TSMC-made ARM chips, Cliff Hou, TSMC Vice President of R&D has said:

If for some reason you just can't wait for the official announcement of the upcoming Trinity platform, then do I have some good news for you. Chinese site EXPreview has supposedly come across AMD presentation slides which detail the upcoming APU. The slides do look to be legitimate as some of the details match what AMD has already released.

AMD has already made public that the new Trinity APU will feature the updated Bulldozer CPU cores dubbed Piledriver. The slides purport that the new cores will process more instructions per clock while leaking less power. The APUs are set to come in dual- and quad-core versions that range from 2.0GHz to 3.8GHz. The TDP appears to top out at 100W.

The lower clock speeds are partially expected due to the fact that the CPU shares its die with a powerful graphics processor. Speaking of the graphics processor, the GPU included on die has up to 384 ALUs which can clock all the way up to 800MHz. The GPU appears that it takes up over half of the silicon die.

Last week, we reported on the Ivy Bridge high temperatures, and whether it was because Intel used TIM instead of solder on the IHS, but now things seem to have changed. A PC EVA forum member has used a Core i7 3770K processor, slapped a Noctua NH-D14 CPU cooler and Prolimatech PK-1 thermal grease, and has tested the chip with and without the IHS on to see if there was a difference with thermal performance.

They used AIDA64 Extreme Edition for idle and load average temperature monitoring, with Prime95 smashing the CPU to generate load. Testing was done at 4.5Ghz with 1.2V on the core. The results?

As we can see, even with the cheaper thermal paste and the IHS layer removed, the cooling performance is relatively unchanged. This also allows a 5-percent margin of error. This is another piece of evidence to show that the heating performance is nothing to do with the IHS, by most likely something to do with Ivy Bridge's revised manufacturing process. This means that an Ivy Bridge should reach lower stable 24/7 clock speeds than a Sandy Bridge chip, but offer it with lower power consumption numbers.

China is looking to define a national standard processor architecture, sources say. If this project is successful, it could be that the new standard would be a requirement in any project that seeks government funding, such as a computer purchases for a school. More important, which architecture will they select?

There are at least 5 architectures that are up for consideration. The Chinese government could also create their own, or extend an existing one. It's somewhat unlikely that they would define their own, especially by committee. You have to realize that a new architecture hasn't been defined here in the West in over two decades.

Officials of China's Ministry of Industry and Information Technology held the initial meeting of the so-called China National Instruction Set Architecture initiative in March. They hosted representatives from about 20 China organizations, which included communications giants Huawei and ZTE as well as a number of academic groups.

Well, well, Samsung have made today quite interesting by announcing the Exynos 4 Quad processor which will be baked into Samsung's next Galaxy smartphone. Samsung have actually come out and revealed this, by saying:

Samsung's new Exynos 4 Quad processor is built on a 32nm process, and hits 1.4GHz and sports over twice the processing power of it's predecessor which is thanks to its High-K Metal Gate (HKMG) low-power technology. We should expect power savings of around 20-percent. Samsung's Senior VP of Product Strategy Team, Hankil Yoon, says:

Nearly every review of Ivy Bridge, including those that were done with engineering samples, has noted that Ivy Bridge runs up to 20*C hotter when overclocked than Sandy Bridge did. People were quick to jump to conclusions on why this was the case, and often these people had nothing to base the conclusions on.

No, these conclusions that people were parroting across the web were wrong. The true answer resides in the fact that, apparently, Intel did not use fluxless solder to attach the IHS (that metal cover over the silicon die) to the Ivy Bridge die. Instead, they have gone back to an older way of doing things and used regular thermal interface material (TIM).

TIM has some major disadvantages to fluxless solder. The biggest, and root cause of this issue is the fact that it doesn't transfer heat nearly as well as fluxless solder. However, it does come with some advantages. You are able to remove the IHS without much risk of damaging the die itself. Could it be that Intel kept extreme overclockers in mind when making this decision? The issue is discussed in more detail at the source below.

ASUS is known for making some incredible products including motherboards. Their internal overclocking team has used an ASUS P8Z77-V DELUXE motherboard along with an Intel i7-3770K to smash 5 world records. This required extreme cooling which came in the way of liquid nitrogen. The records are for benchmark scores and additionally managed 7Ghz on the new Ivy Bridge chip.

The records, as follows, attest to the quality components and engineering that go into ASUS products.

Naming things can be difficult. Think about how hard it was to name your kids. If you don't have kids, think about how hard it was for your parents to name you! A similar crisis plays out when the Intel engineers are trying to come up with the codename for upcoming Intel CPU architectures.

I have to take a moment and apologize for the corny Shakespearean title. It's just that from Intel's perspective, it's true. It doesn't really matter what they call the chip as the name has no influence on the design. The codename just has to be something "nice name that could pass the legal test."

Usually Intel codenames are based on locations in North America, so Sandy Bridge and Ivy Bridge actually break this mold for, what I can tell, the first time ever. To understand how Intel came up with Ivy Bridge, we actually have to look at Sandy Bridge first. Arie Harsat is the Intel engineer who came up with Sandy Bridge.

While the upcoming Ivy Bridge chips (think 3770K, etc) may suck on air, with a little bit of love and some extreme cooling solutions, it looks like these chips could prove to be pretty darn good overclockers. Chinese overclocker x-powerx800pro has managed, with the help of liquid nitrogen (LN2), to achieve a massive overclock of 6.616Ghz!!!! DAMN!

That blows the socks off of my previous record of 5.5Ghz on my old 655K (note: I used a chilled water setup). Using a 63x multiplier and a base clock of 105.03MHz, he was able to achieve the massive overclock of 6.616GHz. Thank you 22nm goodness. However, pay no attention to the reported 'Core Voltage' in the following CPU screenshot.

The core voltage reported is actually a bug. The real core voltage is much higher and comes in at a large, but not ridiculous, 1.85v. This chip is also showing up as an 'ES' which means engineering sample. This means that the retail version could clock much differently. And contrary to the rumor posted previously today, the screenshot does say 77W for the TDP.

With the launch of Intel's Ivy Bridge getting extremely close, it's impossible to keep these retail chips out of the hands of people who haven't signed NDAs. More and more people are getting a hold of the chips and overclocking them. Someone has finally managed to snag a boxed retail version of the chip and found something interesting.

As you can see in the picture above, the rated TDP is 95W. Intel has been telling us along the way that the quad-core Ivy Bridge parts were to be rated at 77W TDP, so this is quite the discrepancy. The only possible reason for this, other than it being the true TDP, is if there was a misprint on a large batch of processors, because sources are saying all of the boxes are marked this way.

Additionally, people overclocking the new part have found it to be a worse overclocker than Sandy Bridge. The main issue with overclocking is that because of Intel's new Tri-gate transistors, they are packed so tightly together that the die size is tiny and doesn't allow adequate heat removal. As such, overclockers are reporting that Ivy Bridge looks to be 200 - 300 Mhz worse than Sandy Bridge.

It looks like we could be in for a treat. Industry sources are saying that Intel will launch the die shrink of Sandy Bridge, dubbed Ivy Bridge, this coming Monday, April 23, 2012. Previously, it had been rumored that Intel was launching the new Ivy Bridge April 29, so it will be coming just under a week early.

This has to be one of the worst launches in recent Intel history. Leaked chips and performance "previews" have been running rampant. Even we have managed to get a hold of one of these chips and put it through its paces. Rumors have been running around that a new, thinner MacBook Pro will be launching with the new Ivy Bridge chips.

The launch will be similar to other Intel launches where the high performance quad-core parts come out first. Then they will be followed by the more power-efficient processors destined for the MacBook Air and Ultrabooks. Intel has said some of these power-efficient chips may launch a few weeks later than originally planned.