While much of the focus from yesterday’s financial call was the poor state of finances, another small tidbit about future plans was mentioned. CEO Lisa Su revealed that AMD has taped out their first FinFET chips back in June this year. While the exact process node has not been revealed, we can assume it is either TSMC 16nm or Samsung/Global Foundries 14nm.
From when the chip first gets taped out to initial production, it takes about a year for the chip to get to market. It takes about 3 months for the chip to get produced, and about 1 month to implement all the fixes. Then the cycle happens once more, taking another 4 months, then finally, production is ramped up, taking a grand total of about 12 months. This means we can expect the FinFET chips to arrive about June/July in 2016, assuming AMD doesn’t hit any major obstacles.
FinFETs will help AMD claw back power efficiency from Intel, who already moved to FinFETs with their mainstream 22nm process back in 2013. Given the timeframe, the chip is likely either a GCN Arctic Islands chip or a Summit Ridge Zen based processor. Both, but especially on the CPU side, can do with better efficiency as Nvidia and Intel are ahead on that front. FinFETs can also reduce overclocking headroom so AMD will have to focus on improving instruction efficiency to compensate. In all, 2016 looks to be a good year for AMD if they can make it there.
We’ve seen and heard a lot about Intel’s mobile and desktop processors lately, but we shouldn’t forget that Intel also creates some of the best server processors in the world. They have just released the latest version of these monsters that come with up to 18 cores and 36 threads, 45MB cache and support for up to 12TB RAM in an eight-socket system.
The new Xeon E7 v3 processors move from the Sandy Bridge architecture to the Haswell but maintains the 22nm manufacturing process for now. We’ll have to wait a little longer for the 14nm versions and while these new chips already impress us, the future process could help with the current power consumption and 165W TDP rating. They are still the best performance per dollar for any business that needs this kind of power.
The new processor family already set 20 new performance world records across a broad range of mission-critical applications where it achieves a 40 percent average performance improvement compared to the prior generation. The processors support configurations with up to 32 sockets and comes with the industry’s largest memory capacity per socket as well as support for both DDR3 and DDR4 memory technology.
New security and reliability capabilities include increased cryptographic performance via the latest Intel Advanced Encryption Standard New Instructions (Intel AES-NI) and Intel Run Sure Technology that is a unique set of reliability, availability, and serviceability (RAS) features for the Intel Xeon processors.
The new Xeon family of processors contains 12 models for various segments including two high-frequency models that are designed for a subset of database applications requiring the fastest cores available. Starting today, the largest system manufacturers will begin announcing systems based on the new Intel Xeon E7 v3 family platform, including Dell, Cisco, HP, Bull, Lenovo, Oracle, Supermicro and many more.
Processors like these aren’t cheap at all and the prices you see below are when purchasing a quantity of 1000, so it’s unlikely that we’ll find many of these in homes around the world in the near future. They’re still some amazing chips and I wouldn’t mind having a couple myself.
The Intel Core i7 5960X, codename Haswell-E, is probably 2014’s worst kept secret. As I am writing this review the full specifications, pricing and pictures of just about every X99 board in existence have already been made public and the NDA is still a few days off. Product launches like this make me wonder what purpose NDAs even serve when they appear to not be worth the paper they are written on. Anyway, politics aside, today we can present your our Intel Core i7 5960X review – at least pretend to be surprised! Intel’s High End Desktop Platform is about to get its first core upgrade since the transition from X48 to X58 when Intel made the leap from 4 to 6 cores, that occurred in 2010. Nearly 4 years later and Intel’s HEDT is making the shift from 6 cores to 8 cores with Haswell-E.
What’s special about Haswell-E apart from the increased core count? Well the X99 platform Haswell-E brings support for DDR4, SATA Express and M.2 (just like Z97 offers), up to 40 PCIe 3.0 lanes and of course 8 core CPUs. If you’re in the market for an upgrade this certainly isn’t going to be cheap, new memory, new storage drives, a new CPU, probably a new power supply…..but I digress. Let’s dive straight into the goodness of the Core i7 5960X. Today we are chucking it on a brand new test system, powered by Gigabyte’s X99 Gaming 5 X99 motherboard and 32GB of Crucial’s fresh-off-the-production line DDR4-2133.
Comparing Intel’s Core i7 5960X to the Core i7 4960X and Core i7 3960X shows some striking similarities. They obviously all share the LGA 2011 package but there are subtle differences. Notably the Core i7 5960X uses a different integrated heat spreader design to the other two.
Moving on over to the rear of the CPU and we actually see a steady decline in the number of built in components. The transition to each newer CPU decreases the number of transistors and other components but we also see an increase in the number of pins. You can see this by comparing the size of the green spacing on the 3960X to the 5960X.
Being a new CPU with a new memory controller this is not compatible with X79 despite still being a LGA 2011 package. Haswell-E takes the LGA 2011-3 package while Sandy Bridge-E and Ivy Bridge-E takes the LGA 2011 package. To prevent people putting the wrong CPUs in the wrong boards Intel has changed the locking points on the CPUs as you can see below.
Intel’s Core i7 5960X comes with a 3GHz base frequency and up to 3.5 GHz with turbo. There’s also native DDR4 support for 2133MHz memory but we are hearing 3000MHz and more is possible with a little bit of tweaking. The other notable thing is a beefy 20MB of shared L3 cache, the most we’ve ever seen on a consumer Intel processor.
Looking at the processor die we can see that it is very different to Haswell for two main reasons: there are 4 more cores and there are no integrated graphics. The new memory controller offers support for only DDR4, there’s no DDR4 and DDR3 combo support like some of our readers may remember on the AMD AM2+ platform.
Intel’s main audience for the Core i7 5XXX series are existing HEDT customers, whether they be X79 or X58. Comparing to X79 Haswell-E and the new X99 chipset brings more cache, more cores, more PCIe lanes, a higher TDP, a different socket, more SATA ports, Thunderbolt support and BCLK overclocking support from the chipset, a feature we also saw moving from Ivy Bridge to Haswell on the mainstream platform.
Like Intel’s previous Extreme Edition CPUs the Core i7 5960X has that $1000 price tag while the Core i7 5930K and 5820K come in for much cheaper. Unlike with Sandy Bridge-E and Ivy Bridge-E going for the 5930K no longer gives you all the performance of the 5960X for less money, the Core i7 5930K has two less cores. The Core i7 5820K also offers less PCIe lanes than the 5930K so each CPU has its own functional purpose: the model separation is better.
Intel released its Devil’s Canyon CPUs at the start of this month in time for this year’s Computex event. Sadly we were delayed in getting our sample of the Core i7 4790K, which means our review has come out a little late, but nevertheless we have Intel’s new flagship on the test bench today for a good thorough review. For those of you who don’t know about Devil’s Canyon, it is Intel’s internal codename for its new Haswell Refresh K series CPUs. Devil’s Canyon includes the Core i7 4790K and Core i5 4690K, both quad core parts based on Haswell Microarchitecture but with speed bumps and a few modifications. There isn’t that much to say specifically about the Core i7 4790K that wasn’t already noted in our Core i7 4770K review, because the microarchitecture is still Haswell. However, there are a few new features that Intel is bringing to the table with Devil’s Canyon that have specifically been done to appease Intel enthusiasts and overclockers. The Core i7 4790K is unique in the regard that Intel have listened to the feedback and concerns of its user base and tried to make specific modifications to eliminate or reduce those concerns.
The first major change comes with regards to the physical characteristics of the CPU. Intel has swapped out the thermal paste under the IHS for a better quality thermal paste which should allow for lower temperatures and better overclocking as a result. Intel have also added additional capacitors to the back of the CPU which they claim smooths power delivery to the CPU die, which again should enhance overclocking potential. Another change that Intel aren’t really marketing is the addition of support for Virtualization Technology for Directed I/O (VT-d) and Transactional Synchronization Extensions New Instructions (TSX-NI)
Intel hasn’t just tweaked the physical design of the CPU either – they’ve also pushed the CPU’s performance even further. For the first time Intel is shipping a consumer processor with a 4GHz clock speed, that’s a barrier AMD broke many years ago but Intel has always been fairly cautious with its clock speeds. The CPU goes even further than 4GHz with a 4.4GHz Turbo frequency, from my experiences on a number of Z97 motherboards this basically means your CPU will always be at 4.4GHz so that’s a significant jump up from anything Intel have ever offered before. Frequency bumps aside the core count, thread count, cache size, graphics and socket all remain the same. Devil’s Canyon CPUs will price match their predecessors on paper but at retailers you can expect to see the older Core i7 4770K and Core i5 4690K slightly cheaper to the tune of $10-50. I also just want to clear up a note about backwards compatibility. Intel’s Core i7 4790K, Core i5 4690K and other Haswell Refresh CPUs will theoretically work in all 8 and 9 series LGA 1150 motherboards. Some motherboard vendors will need to issue BIOS updates to enable this support but all motherboards should get this support because the sockets still have identical pin-outs and the CPU pin-out has not changed either.
Intel announced its successor to the current Bay Trail platform a few months ago – Cherry Trail. Under the new platform the 22nm Silvermont architecture used in the Bay Trail platform gets replaced with the new 14nm Braswell architecture. Like Bay Trail Cherry Trail is low power, cost effective and can run Android, Chrome OS, Linux and Windows operating systems. Braswell SoCs will be available in dual and quad core varieties with 64 bit operation, improved graphics performance, DX11 compatibility and triple screen support.
The latest news from China suggests we will see the new Intel Cherry Trail platform arrive during January 2015. This means the launch is alongside the Consumer Electronics Show where Intel is hoping to push the new platform. Intel is seeking to achieve its long running goal of making entry level Windows and Android tablets even more affordable, it hopes Braswell will offer that,
Intel’s 14nm Broadwell generation of CPUs had previously been on track to launch in Q3 of 2014 but it appears that this may now not happen as VR-Zone reports Intel are experiencing delays. The delays are selective and some parts remain unaffected. Let’s start with the good news which is that the M series and all the desktop parts of the Haswell Refresh will arrive as scheduled. The Z97 chipset is expected to launch in May 2014 while Haswell-refresh CPUs in the desktop and M series will go on sale from mid-April. These CPUs are not affected by the Broadwell delay at all despite earlier speculation. Please note that the Haswell refresh is not the same as Broadwell but in the desktop market they will both share the new Z97/H97 chipset and LGA 1150 socket so there are some similarities. Haswell refresh is still 22nm Haswell but more refined while Broadwell is a reduction in process size from 22nm to 14nm.
Now onto the bad news which is that the Y, U and H series of Broadwell CPUs have all been delayed. The Y series sees the Core i5 5220Y and Core i3 4030Y “Haswell Refresh” CPUs still arriving in Q3 but the Broadwell processors in the series will not arrive until the fourth quarter at the earliest with many more parts arriving in Q1 of 2015. In the U series it is a similar story with most Broadwell parts pushed back to either Q4 of 2014 or Q1 of 2015 but Haswell refresh parts will still arrive this year in Q3 In the H series the Broadwell parts will not arrive until Q1 of 2015 while all the scheduled Haswell refresh parts should launch in Q2 and Q3 of this year.
The delay of the Broadwell U series is expected to cause problems for Apple who will have to delay the release of their new MacBook Air products, though the delay will be segment specific because not all U series parts are delayed by the same amount. For those wondering about H97/Z97 and the Haswell refresh parts for the desktop apparently their isn’t much change over the current desktop Haswell “stuff”. The Haswell refresh parts will mostly just boast a slight speed bump while the H97/Z97 motherboards are rumoured to bring nothing new to the market. As a result VR-Zone expects demand for the new Haswell chipset and refresh parts to be sluggish. In the mobile market it is much of the same for Haswell refresh – mild speed bumps and more mature processors which should give better power consumption.
Intel has officially unveiled its Bay Trail family of mobile SoCs based on the Silvermont CPU micro-architecture. The Atom Z3000 series of Bay Trail-T processors are Intel’s first 22nm processors designed specifically for tablets and mobile handsets. Intel will also reveal Pentium (Bay Trail-M) and Celeron (Bay Trail-D) processors as part of the Bay Trail family. These will be made available for entry “2 in 1s” (notebook/tablets), notebooks, desktops and All-In-One systems. Intel’s Bay Trail is unique in that it is a single hardware configuration designed to support both Windows 8 and Android giving Intel hardware partners more flexibility to create innovative designs.
Intel haven’t disclosed any performance specifics but they claim double the compute performance and triple the graphics performance of its previous generation all with a lower overall power drain and physical size. Intel claim 10 hours of active battery life for its low-power SoC platform (Bay Trail-T) and three weeks of standby time. Intel’s Atom Z3000 series will have four cores, four threads, 2MB of L2 cache and Intel Burst Technology 2.0. Intel says tablets based on the Atom Z3000 series will be available from $199. Intel claims the Z3000 series offerings support for McAfee DeepSAFE Technology, AES hardware full disk encryption, Intel Platform Trust Technology, Intel Identity Protection Technology and Intel Data Protection Technology. Intel says support for 64 bit tablets will come in early 2014.
In terms of Bay Trail-M Intel is offering four SKUs – Intel Pentium N3510 and Intel Celeron N2910, N2810 and N2805 processors. Intel claims devices powered by these SoCs can be as thin as 11mm, be passive/fanless and weigh just 2.2 lbs. Additionally they say expect pricing to start at $199 for clamshell variants, $250 for notebooks and $349 for 2 in 1 devices. Bay Trail-D will be available in three SKUs – Intel Pentium J2850, Intel Celeron J1850 and Intel Celeron J1750.Intel again states these are designed for fanless uses in SFF systems. Systems using these should start from $199.
MSI has just announced that all its X79 motherboards are now ready for Intel’s Ivy Bridge-E processors. Intel’s Ivy Bridge-E retains the same socket and chipset as Sandy Bridge-E so users can just drop the new CPU in an updated motherboard and everything should work just fine. Users will have to update the BIOS of their motherboards first using the latest BIOS versions from MSI’s website, specified below.
MSI claims all consumers with X79 motherboards will have no trouble running these latest processors and overclocking them to their max. Of course these new CPUs are more power efficient so will put less strain on the motherboard’s VRM implementation than the Sandy Bridge-E processors did previously. MSI claims that its Military Class III components combined with its Click BIOS II should give you a flawless overclocking experience. If you’re opting for a Core i7 4960X then you should be able to grab yourself a decent overclock in the 4.4-4.8GHz range depending on the quality of your chip and what voltages you choose.
Ever since Intel released Ivy Bridge to the LGA 1155 platform, LGA 2011 owners were wondering when they would see High End Desktop (HEDT) processors based on the 22nm Ivy Bridge architecture. Up until today the LGA 2011 platform lagged two generations behind the mainstream LGA 115X platforms which are now as far forward as Haswell, two generations ahead of Sandy Bridge. However, today is a great day for all enthusiasts because Intel are taking the covers off Ivy Bridge-E. Ivy Bridge-E brings the 22nm processors to the socket LGA 2011 platform and the X79 chipset. What can we expect to see? Well, similar things to what we saw with the transition from Sandy Bridge to Ivy Bridge except with bigger numbers as we are working with a six core processor not a quad core. Of course there will be quad core processors available and Ivy Bridge-E brings to the market the Core i7 4960X, the Core i7 4930K and the Core i7 4820K which is the quad core while the previous two are hex cores.
Other than the change in architecture there is actually a lot of continuity with Ivy Bridge-E because Intel keep the same socket pin-out, the same chipset and for current LGA 2011 system owners the vast majority of you will be able to keep the same motherboard – all you’ll need is a BIOS update from your chosen motherboard vendor. In today’s review we are going to examine the performance of the new Core i7 4960X in a variety of benchmarks covering gaming, synthetic CPU performance, power consumption and much more. Of course what we are mainly here to try and decipher is whether Intel’s Core i7 4960X is a worthy successor to the Core i7 3960X and if so where does it triumph over its predecessor. What we’ll also be looking for is to see how well the Core i7 4960X stacks up against Intel’s best LGA 115X CPU, the Core i7 4770K, and how well it fares against AMD’s budget Piledriver based eight core the FX-8350.
According to a VR-Zone report Haswell and Broadwell could use significantly different electrical arrangements on the CPU pin contacts . What this basically means is that there will be bi-directional backward compatibility issues with Haswell processors. Haswell CPUs may not work in Intel 9 Series motherboards (Z97/H91/B95…etc) and Broadwell CPUs may not work in Intel 8 Series (Z87/H81/B85…etc) motherboards. The reason for this is explained by VR-Zone as:
“The most notable changes are in the V_PROC_IO connection, as it requires a 1.05V power source, VCCST, because of a new type of power supply required, and to THRMTRIP because of a slightly different chip topology.”
What this means is that only motherboards which support the Haswell “refresh” chips will support Broadwell CPUs. We are currently on the Haswell generation of CPUs, there will be a new refreshed generation of CPUs based on Haswell architecture shrinking down from the 22nm process to the 14nm process and then there will be Broadwell which will use the 14nm process but use a redesigned architecture. If VR-Zone are correct then all three CPU generations will share the LGA 1150 socket but only Haswell Refresh and Broadwell CPUs will work together on their respective motherboards and the current Haswell generation will only work on current Haswell motherboards.
VR-Zone’s sources also state that there is mixed information on whether Broadwell has DDR4 support or not. It is confirmed that the Intel 9 Series chipset has SATA Express support.
According to some information from VR-Zone Intel’s Ivy Bridge-E platform will be quite different to Sandy Bridge-E. The Ivy Bridge-E CPUs will be native 6 core dies as opposed to cut down 8 core dies like the Sandy Bridge-E CPUs were. While this may seem like a minor or insignificant change, it is not. It does indeed feature the same 15MB of cache and same six cores as Sandy Bridge-E but the reduced die size should improve the CPU in a couple of ways. Firstly it should consume less power, as we have already seen from early reviews, and secondly, it should give more overclocking headroom both on the memory controller and on the CPU itself
Unlike the current Ivy Bridge chips the Ivy Bridge-E chips have no integrated graphics and (should) come with a soldered heatspreader, not the crappy thermal paste Intel uses in Ivy Bridge and Haswell Core series processors. These two things mean overclocking headroom will be greatly improved. Graphics enthusiasts will also be pleased with the true PCIe generation 3.0 certification compared to the hybrid 2.0/3.0 certification we saw on the current Sandy Bridge-E CPUs. Finally, Intel is also reported to have successfully matured the 22nm FinFET process to solve some leakage problems seen on other 22nm CPUs that can hinder performance slightly.
If you are interested in getting a Ivy Bridge-E CPU when they come out later this year, expected September, then you may be pleased to know pricing details have surfaced from VR-Zone. The first “entry” model, the Core i7 4820K, comes with a price tag of $310 USD which makes it $5 more expensive than its predecessor though it does come with a fully unlocked multiplier and a higher base clock. The Core i7 4930K, comes with a price point $39 USD lower than its predecessor despite being a better overall part. While the Core i7 4960X boasts a price about $69 USD lower than its predecessor which is welcome news given the extortionate three figure price of the Core i7 3960X.
Despite lower launch prices than predecessors Ivy Bridge-E is still a hugely expensive platform and again we will probably see the Core i7 4930K being the favourite part, much like the Core i7 3930K, because it offers nearly all the features of the much more expensive X part except with 3MB less cache and a marginally slower base clock which can be reached by overclocking.
Will you be buying Ivy Bridge-E or is it too expensive for your budget?
Image #1 courtesy of Intel and Image #2 courtesy of VR-Zone
Intel is planning on releasing a brand new Haswell CPU called the Core i7 4771. This will be a locked multiplier Core i7 Haswell processor with identical clock speeds and design to the Core i7 4770K. That means a 3.5GHz base clock and 3.9GHz turbo clock across four cores and eight threads. 8MB of cache, a 22nm Lithography and an 84W TDP top off the rest of the specifications for the Core i7 4771. It will come in to replace the Core i7 4770 which runs at 3.4GHz base clock and 3.9GHz turbo clock.
VR-Zone have benchmarked the new Intel CPU and unsurprisingly it performs exactly like a Core i7 4770K except without overclocking potential. The Core i7 4771 will probably replace the Core i7 4770 in the long term and sit alongside the Core i7 4770K as the leading Haswell Core i7 processors. Expect a slightly higher price premium of about $10 compared to the current Core i7 4770.
According to Hermitage Akihabara, via TechPowerUp, Intel’s Ivy Bridge-E processors will arrive in just under 2 months on September 11th. These new processors will succeed the Sandy Bridge-E line up but will work on the same socket and X79 chipset with a BIOS update. New revisions of X79 motherboards with BIOS updates are likely to be issued for the release by most vendors. With the launch we will see the Intel Core i7 4960X, 4930K and 4820K all launched.
The i7 4820K has four cores at 3.7GHz, a turbo frequency of 3.9GHz, 10MB of shared L3 cache and is partially overclockable. The i7 4930K has an unlocked multiplier, a base speed of 3.4GHz, a turbo of 3.9GHz and 12MB of shared L3 cache. Finally the i7 4960X also features an unlocked multiplier, 3.6GHz base clock speed, 4GHz turbo and 15 MB of shared L3 cache. I would also go as far to say that we will probably see an i7 4970X in Q2 or Q3 of 2014 which will have an unlocked multiplier, 3.8GHz base speed and 4.1GHz turbo plus the 15MB of shared L3 cache.
The new Ivy Bridge-E processors could also bring a new chipset or chipset revision featuring more SATA III 6Gbps ports and integrated USB 3.0 but this is still unconfirmed. If you missed our news story on the performance (p)review of the Intel Core i7 4960X then be sure to check that out here.
Tom’s Hardware are at it once again with their Intel Core i7 4960X preview coming out way before the Intel NDA. While media publications line up to complain to Intel over this blatant violation of industry rules, for our readers this is still a great opportunity to check out the performance of Intel’s upcoming flagship CPU of the flagship LGA 2011 consumer platform.
The battery of tests they have put the Core i7 4960X through are mainly productivity and work focused so there isn’t much to show you in the way of gaming. Though that said the first test is of 3DMark11 and you can see a greatly improved physics score from the Core i7 4960X showing the generational performance boost. Note however that the Core i7 4960X does have an extra 0.1GHz over the Core i7 3970X which accounts for some of the performance boost.
Memory bandwidth is marginally improved over the previous generation of Sandy Bridge-E CPUs.
Most other traditional synthetic benchmarks show marginal increases for the Core i7 4960X over the previous generation.
The main design win of the Ivy Bridge-E Core i7 4960X is power consumption. As you can see it is significantly more power efficient that its predecessor the Core i7 3970X.
Average power consumption gives this CPU more resemblance to a Core i7 2700K. Power effiency really is excellent and leaves competing solutions trailing in the distance.
Intel have not officially commented on the review though Intel staff have done so off the record. As usual most Intel staff are saying the review has lots of inaccuracies and inconsistencies so take these results with a pinch of salt.
Despite Haswell now being Intel’s latest consumer platform, Intel is still looking to keep Ivy Bridge and LGA 1155 alive with some more CPU releases. CPU World have stumbled across four new CPUs that Intel will be releasing in the near future. They include two Core i5 and two Celeron models.
All four CPUs do not feature hyperthreading but all feature similarly clocked HD Graphics. The two Celeron parts are both dual cores while one is a “T” model meaning a reduced TDP for energy efficiency. This has a slightly lower clock speed but they both have identical cache, graphics and memory support. The next two parts are both Core i5 parts with four cores and HD 2500 Graphics. Again these have identical cache and memory support with the only difference being that the “S” variant has a lower TDP.
These releases are designed to keep Ivy Bridge bolstered in that entry/mid-range segment since Haswell is currently only high-end models – we are yet to see Haswell Core i3 and below. That said Intel will probably retain Ivy Bridge support for a while to come yet since LGA 1155 was a big selling platform for Intel especially since it lasted two generations of CPU and motherboard releases.
Image #1 courtesy of Intel and image #2 courtesy of CPU-World
Intel’s Haswell has been out for a while now but there hasn’t been any sign of any sub Core i5 desktop Haswell processors yet. That said these Core i3 units still have a while to go but we can get some kind of estimation of their performance level as benchmarks of a dual core hyper threaded Core i5 4570T have been revealed. With two cores it mirrors what we will probably expect to see from a Haswell Core i3 as most of these are dual core hyper-threaded parts. You can see the specifications of the Core i5 4570T below in the graph. The T moniker denotes the energy efficiency of the unit and this is reflected by the 35W TDP.
In terms of performance it is impressive and performs better than a lot of AMD counterparts with much higher clock speeds and power consumptions. It also delivers performance close to that of Sandy Bridge “K” favourites like the 2500K and 2600K processors. In Battlefield 3 it did particularly well.
In The Elder Scrolls V Skyrim it did equally well showing that less cores doesn’t equal less performance. It dispatched of AMD’s FX-8350 8 core processor and blew away the best Ivy Bridge Core i3 dual core part from the last generation, the i3 3220, which runs at 3.3GHz with 2 cores and four threads while this runs at 2.9GHz with the same physical core/thread count.
Though it does consume a fair bit more power than the Ivy Bridge dual core part though the performance is still very impressive for the tight 35W TDP envelope. We can see that the Haswell architecture clearly scales very well at lower clock speeds.
If you are interested in seeing more graphs from the dual core Haswell Core i5 4570T review then be sure to check out the source article here – PC Games Hardware (de)
Intel’s NUC has spawned a new generation of mini PCs. We recently took a look at the Next Unit of Computing (NUC) from Intel and we found it to be a very impressive unit but was marginally let down by components lacking performance, a high price tag and a missing wireless module. Gigabyte’s BRIX looks set to take on two of those major flaws in Intel’s new reference NUC platform as they have opted for a Core i5 and have included a wireless module.
Today we are reviewing the Gigabyte GB-XM11-3337 BRIX which is like Intel’s NUC except Gigabyte have redesigned the chassis, changed the I/O, changed the BIOS and opted for some higher spec hardware. While Intel’s NUC measures in at 1.55 inches thick Gigabyte’s BRIX is much smaller at 1.18 inches thick. In fact the Gigabyte BRIX is just a generally more compact unit than Intel’s NUC. Furthermore the wireless module is included as standard and USB 3.0 has been added too. Finally there is a DisplayPort and HDMI instead of two HDMI meaning you have slightly more flexibility when it comes to display outputs.
Additionally the main change comes in the form of the CPU. The Intel NUC we reviewed had a dual core 1.1GHz 32nm Sandy Bridge based CPU whereas this Gigabyte BRIX uses an Ivy Bridge 22nm i5 3337U processor that has 2 cores and four threads at 1.8GHz stock and 2.7GHz turbo. It also supports more memory bandwidth at up to 1600MHz compared to the 1333MHz on the Intel NUC we looked at that had a Celeron 847. Finally it also boasts Intel HD 4000 graphics which aren’t going to play any games that well but are still a huge leap forward over the graphics part of the Sandy Bridge ULV processors like the Celeron 847. However, this comes at a cost as the Core i5 3337U has a tray price of $225 which is more than the entire Intel Celeron 847 based NUC system costs – straight away we know this isn’t going to be cheap but let’s see how much performance this device offers.
Below you can see the full specifications of Gigabyte’s BRIX GB-XM11-3337 system and without any further ado we will now proceed through the rest of this review.
Intel’s Haswell has been out for a while now but there are still some unreleased models we are expecting. According to a VR-Zone story the rest of the Haswell line up will arrive on September the 1st. This will consist of a single Core i7 model, two Core i5 models, five Core i3 models and five Pentium models. There are also a pair of Ivy Bridge Core i5 processors and a trio of Ivy Bridge based Celeron processors coming too on September the first.
With regards to the specifications of the Haswell processors WCCFTech have created this excellent summary table that shows you all you need to know about these upcoming Intel Haswell processors. Interestingly enough most of the processors exceed the 3GHz mark quite comfortably which is something that we haven’t seen for a while with Intel processors. All Core i3 models have hyper threading and HD 4600 graphics while there are a couple of T models that have reduced 35W TDPs over the 54W TDP of the standard Core i3 models. The new Core i7 4771 is simply an i7 4770K that is locked and it has an 84W TDP.
Image #1 courtesy of VR-Zone and Image #2 courtesy of WCCFTech
The guys over at VR-Zone have a long tradition of getting leaks about new Intel platforms quite early and they have published news of Intel’s new Haswell-E platform before we’ve even seen its predecessor Ivy Bridge-E. According to their reports the new Haswell-E platform will hit the market in the second half of 2014. The new Haswell-E CPUs will have 6 or 8 cores, more than the currently available 6 on the Sandy Bridge-E and Ivy Bridge-E platform.
Furthermore these CPUs will boast 20MB of L3 Cache and 16 threads. The TDP may rise slightly to 140W up from 130W, on selected parts, as two extra cores means extra power required despite generational power savings from moving to Haswell. The roadmap also crushes old speculation that 14nm processors would arrive in 2014, as they now appear set for 2015 at the earliest. Interestingly there will be no more quad core parts in this platform like the i7 3820 or i7 4820K as Intel is only offering 6 and 8 core parts with Haswell-E.
The Haswell-E platform itself (Wellsburg) can manage six USB 3.0 ports, eight USB 2.0 ports and ten SATA III 6Gbps ports. There are also 40 PCIe lanes, the same as Ivy Bridge-E and Sandy Bridge-E, and these are all at Generation 3 speeds and are controlled on the processor. The system will support quad channel DDR4 2133MHz memory natively.
Finally performance will be at least 33% better (due to the extra cores) and then if you add generational improvements in the architectures we could see up to 50% more performance over Ivy Bridge-E which is already 10% faster than Sandy Bridge-E. The socket will retail LGA 2011 layout and pins but will be “LGA 2011-3” which has a more efficient design, better package handling and comes with additional wings.
Intel’s Xeon Phi co-processors are aimed at high density servers and their current line up includes just the 5110P and 3100. This range of processors from Intel is set to be expanded this month as Intel adds some higher end models and some more budget models with 3100 and 7100 models plus an additional 5100 model.
The Xeon Phi 3100 series will be positioned as the “inexpensive” (for businesses) co-processors for compute-bound tasks. Intel is launching the 3120A and 3120P for this, both are identical but feature different cooling solutions. These PCI Express add in cards each feature 57 x86 cores at 1.1GHz, with 28.5MB of L2 cache, 6GB of onboard cache, a 300W TDP and 1003 GFLOPS of performance.
The Xeon Phi 5100 series will add the 5120D which boasts 60 cores at 1.05GHz with a slightly lower TDP of 245W. This unit offers 1073 GFLOPS of performance.
The Xeon Phi 7100 series adds the 7120P and 7120X models which have 61 cores at 1.25 GHz. The difference between the two come in the form of the cooling solution and connector type
Full specifications can be seen below:
For those of you who are unaware as to what “Xeon Phi co-processors” are, they are Intel’s attempt at competing with AMD and Nvidia in offering add-in cards that can provide some serious compute performance. AMD and Nvidia already provide these in the form of the AMD FirePro, Nvidia Quadro and Nvidia Tesla enterprise add-in cards. However, about 9 months ago it emerged that Intel’s Xeon Phi could barely compete with AMD and Nvidia’s last gen graphics architecture design in terms of compute performance. I have yet to see a current generation GPU architecture vs Intel Xeon Phi comparison but you can bet that the current generation GPU architecture wins comfortably and at a lower price. That said GPU compute is likely to be the way forward in terms of cost-efficiency but Intel’s Xeon Phi does perform admirably and Intel is pumping in a lot of money to it.
A new roadmap of Intel products has been leaked and it shows details on the distant future release of the Intel HEDT Haswell-E platform. Apparently the new Haswell-E “Lituya Bay” platform, which will succeed the yet-to-be-released Ivy Bridge-E platform, will bring the first consumer DDR4 support.
SK Hynix and Micron discussed the possibility of bringing DDR4 to the Haswell-E platform at Intel’s Developer Forum 2013 event. This would make sense as we saw companies like Crucial showing off their DDR4 modules at this years Consumer Electronics Show (CES) event in Las Vegas. However, it is to be expected that DDR4 will arrive in the server and business market much earlier than in the consumer market.
That said, apparently it was revealed at IDF 2013 that Intel’s Haswell-E HEDT platform would be the first to bring DDR4 memory support with a DDR4 IMC (integrated memory controller). The latest roadmap would suggest we will see Haswell-E at the end of 2014 and early 2015 meaning that Ivy Bridge-E will only have a 1 year “shelf-life”.
It is expected Haswell-E will retain LGA 2011 socket compatibility and will still be based off 22nm technology, although the above roadmap is misleading. We will see the 14nm Skylake platform launch after Haswell-E in early 2015. Both Haswell-E and Skylake are listed as LGA socket processors meaning there is no danger of Intel exiting the DIY CPU market any time soon like some had previously anticipated.
What are your thoughts on Haswell-E bringing DDR4 support and arriving in late 2014?
Intel’s Chief Executive Office (CEO), Paul Otellini, has been revealing some potentially excellent news at Intel’s first quarter earnings call. Speaking about Intel’s upcoming new Atom processor release Otellini hinted that we could see huge drops in prices on Windows 8 devices such notebooks. The latest Intel Atom Bay Trail processors could allow Windows 8 touch based devices to sink as low as $200 per device – an incredibly impressive feat. Traditionally that kind of price point has only been occupied by ARM powered Android tablets and really low-spec netbooks.
Intel’s Bay Trail is being dubbed the most powerful Atom processor ever, as you’d expect from the latest generation hardware. It brings an entirely redesigned architecture and shrunk manufacturing process of 22nm, compared to 32nm. If Intel can bring these Bay Trail chips to market in a very cost effective way then we could see Windows 8 devices recover because currently what is holding a lot of people back is the high price.
“If you look at touch-enabled Intel-based notebooks that are ultra-thin using [Bay Trail] processors. Those prices are going to be down to as low as $200… Bay Trail is going to be a great product in that segment of the market and enable stunning performance relative to what the competition can bring.”
What are your thoughts on potentially being able to pick up some Windows 8 devices for as low as $200? Might it encourage you to finally buy a Windows 8 device?
Intel has been investing serious amounts of money into developing its Atom processors for the mobile market for quite some time now. Intel’s ValleyView based Atom processors, based on the Bay Trail platform, will feature the latest and greatest in Intel manufacturing technology, that is 22nm processing.
According to Industry Sources, probably from within Intel, the new Intel Atom quad core processors based on the ValleyView design will be competitive with their ARM tablet counterparts currently on the market. The 22nm Intel Silvermont cores feature much faster execution rates, higher clock speeds, better memory bandwidth, dramatically improved graphics and lowered power consumption.
Intel is expecting its quad core ValleyView Atoms to provide solid competition against ARM in the tablet, netbook and nettop markets. While Intel is also looking to gain within the server space by producing 8 core variants of their ValleyView processor for high performance low power variants that can compete with ARM based servers.
Intel is looking to prove everyone wrong as the general word in the market is still that ARM is more cost-effective and better performing than equivalent Intel Atom SoCs. Intel will be hoping to claw back some ground to its strong ARM competitors such as Qualcomm’s Snapdragon 800 and Nvidia’s Tegra 4. Currently you will be hard-pressed to find Intel’s Atom processors in many smartphones or tablets but Intel is looking for that to change with ValleyView.
What are your thoughts on ValleyView? Would you buy an Intel powered tablet or smartphone?
Intel’s Haswell, LGA 1150, platform is expected to hit the market in June 2013 and by then the mainstream platform will be two generations ahead of Intel’s High End Desktop (HEDT) platform which is still stuck on Sandy Bridge architecture. However, Intel will be bringing the HEDT platform, socket LGA 2011, up to Ivy Bridge level in Q3 of this year.
The new portfolio of Ivy Bridge-E products looks very similar to the current generation except with an advancement in the numerical naming strategy. In fact, all the new Ivy Bridge-E processors will be fully compatible with the current LGA 2011 packagae and X79 motherboards.
The current line-up consists of the i7 3820, i7 3930K, i7 3960X and i7 3970X. The new line-up is looking similar with an i7 4820, i7 4930K and i7 4960X with an i7 4970X probably expected about 6 months after the first wave launches.
The i7 4820 has four cores at 3.7GHz, a turbo frequency of 3.9GHz, 10MB of shared L3 cache and is partially overclockable. The i7 4930K has an unlocked multiplier, a base speed of 3.4GHz, a turbo of 3.9GHz and 12MB of shared L3 cache. Finally the i7 4960X also features an unlocked multiplier, 3.6GHz base clock speed, 4GHz turbo and 15 MB of shared L3 cache. I would also go as far to say that we will probably see an i7 4970X in Q2 or Q3 of 2014 which will have an unlocked multiplier, 3.8GHz base speed and 4.1GHz turbo plus the 15MB of shared L3 cache.
Since these parts are all based on the Ivy Bridge process they will be 22nm parts and the TDP for all three SKUs is expected to be 130W. Availability is some time in Q3 2013.
Intel brought the 22nm architecture to the desktop in the form of “Ivy Bridge” back on April the 29th 2012. The transition to 22nm represented a large step forward in power efficiency and a modest step forward in clock-per-clock performance. Now Intel is promising that this transition will hit their Atom processors by the end of this year and this is particularly exciting news for mobile and tablet users. Intel Atom processors are designed from the ground up with power consumption in mind, they have to be incredibly power efficient for the sake of battery life and that often means Intel have to reduce clock speeds to get the right balance. However, the introduction of 22nm means that Intel can now hit these same power envelopes with much more performance. As well as being able to offer lower-powered options to really squeeze the absolute maximum battery life out of these 22nm tri-gate Atom System-on-Chips (SoCs).
However, as we wait for these new 22nm Atom SoCs to be released, Intel has just announced three new SoCs based the 32nm Clover Trail+ Atom micro-architecture at the Mobile World Congress (MWC) event in Spain. The Intel Atom Z2580, Z2560 and Z2520 are the three new SoCs to hit the market.
Intel are claiming double the compute performance and three times the graphical capabilities over the previous generation of mobile Atom processors – all while maintaining “competitive low power”. New features include support for 16 Megapixel (MP) image sensors, burst mode capabilities capturing 8MP images at 15 frames per second (FPS), hardware acceleration for 1080p video encoding and decoding at 30 FPS and enhanced High Dynamic Range (HDR) capabilities – something Intel needed to include because Nvidia’s Tegra 4 is also promising this. However, integrated LTE support didn’t materialise although Intel does offer the 4G capable XMM 7160 modem. As an extra Intel is also providing real-time facial recognition technology as part of the Clover Trail+ platform.
Lenovo’s IdeaPhone K900 will be the first phone to boast Intel’s flagship Z2580 Atom SoC on its 5.5″ high resolution display. More products using these SoCs have been promised by ASUS and ZTE, while Intel claims Android tablets will be powered by these in the not-so-distant future.
The main story from the even though is Intel’s Merrfield 22nm Atom SoCs which Intel will start producing this year. If ARM progress is anything to go by, Intel are really going to need the lowered power and improved performance that the 22nm process has to offer. Add to that the fact Intel is already having a hard time encouraging App and OS developers to opt for x86 instruction set support – the pathway to mobile success for Intel looks long and obstacle filled.
Production begins in late 2013 meaning finished products will not arrive until Q2 of 2014. By then Nvidia’s Tegra 4 and Samsung’s Exynos 5 Octa (eight core) SoCs will already be established in the market. Best of luck Intel – you’re going to need it.