Click on the competition images to go straight to the competition page, or click here for a more detailed overview at HWBOT.
World Tour 2017 and HWBOT X
Road to Pro 2017
|XTU||Core i7 7820X||5856 MHz||rsannino||3702 marks||116.5 pts||0 0|
|Unigine Heaven - Xtreme||GeForce GTX 1080 Ti||2610/1610 MHz||OGS||11207.26 DX11 Marks||103.7 pts||0 0|
|HWBOT x265 Benchmark - 1080p||Core i7 7820X||5785 MHz||rsannino||91.18 fps||80.5 pts||0 0|
|Cinebench - R15||Core i9 7900X||5960 MHz||sofos1990||3327 cb||60.6 pts||0 0|
|Cinebench - R11.5||Core i9 7900X||5960 MHz||sofos1990||35.87 points||54.4 pts||0 0|
|XTU||Core i7 7800X||5000 MHz||aerotracks||2503 marks||54.2 pts||0 0|
|XTU||Core i5 6500||3400 MHz||miker2ka||1110 marks||49.9 pts||0 0|
|Cinebench - R11.5||Core i7 7820X||5886 MHz||rsannino||28.49 points||46.3 pts||0 0|
|HWBOT x265 Benchmark - 1080p||Core i7 7800X||4900 MHz||aerotracks||57.75 fps||43.9 pts||0 0|
|Cinebench - R15||Core i7 7820X||5899 MHz||rsannino||2638 cb||36.7 pts||0 0|
Click on the competition images to go straight to the competition page, or click here for a more detailed overview at HWBOT.
Our Motherboard Memory Lane series today arrives at the AMD Socket FM1 era. The arrival of the FM1 Socket heralded a significant change in direction for AMD which launched its first Accelerated Processor Units or APUs in the market. Aimed at the mainstream to entry-level segment the new platform hoped to woo PC enthusiasts and overclockers with a relatively decent CPU coupled with a much beefier integrated GPU. Let’s take a closer at the new platform, the motherboards and processors that were popular during this era and of course, some of the most notable scores posted on HWBOT.
The arrival of the AMD FM1 Socket marked a pivotal change in the overall AMD product lineup. Socket FM1 would become the mid-range and entry-level platform leaving the mature AM3+ platform to spearhead its high-end offerings. Whereas previous mainstream platforms from AMD had relied upon a Northbridge Chipset such as the AMD 880G and AMD 880GX to deliver integrated graphics and digital display outputs, the new FM1 platform used Accelerated Processor Units had a much more substantial GPU baked into the processor itself. AMD would later release its Bulldozer-based AMD FX series processors on the AM3+ platform in an attempt to better compete with Intel’s recently arrived Sandy Bridge offerings.
This week our Motherboard memory Lane series turns its attention to the AMD AM3 Socket. The platform boasted an updated processor series that brought DDR3 memory support to AMD platforms for the first time, plus a broad range of dual-core, tri-core, quad-core and hexa-core models that AMD hoped would woo the hearts and minds of overclockers on HWBOT. Let’s move on and check out the motherboards, chipsets, processors and scores that defined the AMD AM3 generation.
With Socket AM3 we have the arrival of a new and updated series of AMD Phenom II processors. The most notable feature of the new chips was the modified memory controller (residing in the CPU) which now supported DDR3. This was AMD’s first stint at supporting DDR3 memory which had by this stage become reasonably affordable compared to DDR2 modules, largely thanks to Intel again making the move much earlier to help drive adoption. At launch AMD Phenom II X4 and X3 processors based on the Denab iteration of the K10 architecture arrived supporting dual-channel DDR3-1333 memory with multipliers available for frequencies as high as DDR3-1600. One other key difference was a larger L3 cache and the use of a 2GHz HyperTransport bus (compared to 1.8GHz on the previous gen).
The subject of this week’s Motherboard Memory Lane article is the AMD AM2+ platform. Strictly speaking the AMD AM2+ socket is historically the successor to the Socket AM2 and the predecessor to Socket AM3. The AMD AM2+ Socket was launched alongside the company’s first true quad-core and tri-core processors; the AMD Phenom series. Let’s take a look at the platform itself, the processors that it supported, the boards that were popular and of course the scores that were made by HWBOT members at that time.
After the roaring success of its K7 Athlon architecture CPUs and its follow up, the K8 Hammer architecture which brought us the first 64-bit, dual-core processors, the K10 architecture (technically referred to as the AMD 10h Family) arrived with a new Phenom brand name and the company’s first true (monolithic) quad-core processor series. Sounds pretty exciting, but in fact the new platform was received by tech media and enthusiasts with some real disappointment. Let’s look at why this happened.
Clock speeds were lower than expected, the platform remain limited (initially at least) to DDR2 memory and suffered a from translation lookaside buffer (TLB) bug that could cause a system lock-up (in fairly rare circumstances). Perhaps even worst of all, the new AMD Phenom chips simply could not keep up with Intel’s performance. You could almost point to the AM2+ launch as the beginning of the company’s drift into the void of non competitiveness. A void from which it is only now, ten years later, beginning to return, thanks to its new Zen architecture offerings.
Welcome to the latest edition of our Motherboard Memory lane series here on HWBOT. This week we turn our attention to the AMD AM2 platform, a platform that most notably featured an updated integrated memory controller that supported DDR2 standard memory. The platform also arrived with a new series of AMD 64 X2 processors based on a new and revamped K8 architecture. Let’s take a look at the AMD AM2 platform, the boards and processors that were popular with overclockers at that time and some of the outstanding scores that were submitted to HWBOT.
The AMD AM2 platform officially arrived in May 2006 and was the direct replacement for Socket 939. Although physically the AM2 Socket used exactly 940 pins in the same ZFI (zero insertion force) socket design that is used today, the new platform did not physically support previous generation Socket 940 CPUs due to an intentionally incompatible pin layout. The new socket however did debut a different heatsink retention mechanism with a cage-like design that was attached to the motherboard using four screws, not two. The heatsink / cooler dimensions remained unchanged however.
The new platform arrived with a range of single and dual-core processors, initially based on the Windsor (dual-core) and Manilla (single-core) architectures which were members of the original K8 family which had debuted several years earlier with Socket 754. Subsequent platform refreshes added Brisbane and Orleans architecture models.
Today in our Motherboard Memory Lane series we take on the classic AMD Socket 939 platform. After the heady heyday of the Socket A era, Socket 939 saw AMD build on the 64-bit architecture processors that debut on Socket 754, adding dual channel memory support, improved overclocking and eventually dual-core models. All of which makes it a memorable platform for many overclockers, especially when you thrown in some truly magnificent motherboards from DFI. Come with us as we recall the chipsets and boards that defined the era, plus a few of the scores that were submitted to HWBOT at the time.
The AMD Socket 939 platform arrived on the market in June 2004, just nine months after the company launched its Socket 754 platform. As with Socket 754, the new platform was designed to support the latest Athlon 64 and Sempron processors, eventually going to support Athlon 64 FX and Athlon 64 X2 models. The actual Socket 939 design was in fact very similar to Socket 940 (just one pin less) which was essentially AMD’s server platform. Socket 940 supported Opteron and Athlon 64 FX chips which required registered DDR memory. In terms of design Socket 939 was launched as consumer grade platform and featured a dual memory controller and support for more affordable and readily available non-registered memory modules.
At launch the new platform arrived with a new range of Athlon 64 processors, and as with the previous Socket 754 platform, two main chipset options; VIA and Nvidia. At launch the VIA K8T800 competed against the Nvidia nForce3 platform. Eventually the most popular choice with overclockers became the Nvidia nForce4 chipset, a single chip solution that evolved to offer a 1GHz HyperTransport support and SLI support, one of the first AMD platforms to do so. While the standard nForce4 Ultra MCP (Media Communications Processor) offered 20 lanes of PCIe the nForce4 SLI packed 38 lanes of PCIe (which had now replaced the aging AGP interface). The nForce4 Ultra chipset also offered 10 USB 2.0 ports, 2x IDE ports and 4x SATA 3Gbps ports, Gigabit LAN and AC’97 2.3 audio.
EVGA have just announced a really cool event where enthusiasts who purchase the latest EVGA GeForce GTX 1080 Ti K|NGP|N Edition graphics card will have the chance to win the ultimate prize; a two day benching sessions with the man himself, Vince "K|NGP|N" Lucido. Inside each KP Edition 1080 Ti card retail box, users will find a K|NGP|N Golden Card which can be used to become one of six lucky winners. Each winner will be flown to the EVGA headquarters in Taiwan where they will be treated to three nights accommodation with meals included, a chance to meet and do some extreme overclocking with Vince, plus the opportunity to win a very special prize.
Calling all EVGA GeForce GTX 1080 Ti K|NGP|N owners! EVGA is giving six (6) lucky winners the exclusive opportunity to meet K|NGP|N himself. The winners will fly to Taiwan to meet and overclock with the king guru himself, Vince "K|NGP|N" Lucido. EVGA will fly you to Taiwan, put you up in a hotel, and provide meals for the duration of the trip. The trip will feature a two (2) day extreme overclocking workshop with K|NGP|N...and a special prize!
Winners will be chosen by the end of September 2017 with the event penciled in to take place on November 13th. It’s great to see EVGA put together this event and create engagement with end-users and the guys at EVGA that make all the OC magic happen. Really good to see. You can learn more about the event, including the full ‘Terms and Conditions’ here on the EVGA website.
The latest edition of the OC Show is available now on Overclocking-TV. Ciro, Toolius and Buildzoid broadcast Episode 4 of Season 4 just a few days ago live on Twitch-TV, once again discussing the latest issues that emerged in the world of overclocking and the enthusiast PC market as a whole.
The show kicks off with a chat about the recent exploits of overclockers Xtreme Addict (Poland) and Sofos1990 (Greece) who have been competing on the Cinebench R15 and R11.5 benchmarks. XA uploaded his new Global First Place scores, displacing scores from Sofos from the top of the table, only for Sofos to respond within minutes by uploading a higher score that he had prepared for exactly that eventually. The practice of keeping backup scores on hand for an immediate response is known as sandbagging, a topic which goes on to dominate the discussion for the next few minutes.
The guys then turn their attentions to the contests happening on OC-ESPORTS where the GIGABYTE Beat the Heat contest is heating up, as well as the Rookie Rumble #46 contest which is currently dominated by a group of US-based Rookies. Other topics that are discussed in some detail is the release of the Intel Core i9 7920X and its lower than expected clock frequencies and the reception of the X299 platform and the prospect of some pretty hot chips that may not actually prove to be too useful at overclocking. The other major topic of debate is the upcoming launch of AMD’s Threadripper platform, another HEDT grade platform that is poised for launch. The guys take a look at leaked images of TR4-platform boards from ASUS and GIGABYTE, discussing the VRM cooling and other design features. This is followed by an interesting discussion about what to expect from Threadripper and also AMD Vega GPUs when they arrive in the next few weeks and months.
You can find the full OC Show from the OC-TV team here on the OverClocking-TV YouTube channel.
With the recent launch of Intel’s re-jigged HEDT processor lineup and the arrival of the X-series Core i7 and i9 model CPUs, it seems apt that this week’s Throwback Thursday is all about the arrival of another HEDT series - Sandy Bridge-E. Back in July 2011 the aging Gulftown platform was really beginning to creak, which is why so many of us were looking forward to getting hands on with an entirely new series of high-end parts based around the ground breaking Sandy Bridge architecture. Early rumors and leaks floating around certainly seemed to indicate that the new X79 platform would be a beast in pure performance terms. Here’s what we wrote on July 22nd 2011:
Slides of a key presentation to Intel's partners was leaked to sections of the media, which reveal Intel's own performance testing of the Core i7-3960X Extreme Edition, the top-model of the socket LGA2011 "Sandy Bridge-E" processor series. Meet the family here. In its comparison, Intel maintained the Core i7-990X Extreme Edition socket LGA1366 processor as this generation's top offering.
From these test results, the Core i7-3960X Extreme Edition is pitched to be about 47.25% faster on average, compared to Core i7-990X Extreme Edition. Intel is attributing the performance boost, apart from the normal IPC increase, to the 33% higher bandwidth thanks to the quad-channel DDR3 IMC, and the new AVX instruction set that accelerates math-heavy tasks such as encoding.
Greek Overclocker OGS clearly had a busy and productive couple of days over the weekend, managing to post a bunch of really impressive scores that have ultimately pushed him up to fourth place on the HWBOT worldwide rankings. Among the nine submissions that he posted yesterday, we have four Global First Place scores. Let’s take a look at these scores and the fairly broad range of hardware used to achieve them.
Let’s start with OGS’s skills at 3D benching and take a look at the new single GPU Global First Place score for the 3DMark11 Performance benchmark. The new highest score for a single card now stands at 44,282 marks, which beats the next best from OC legend k|ngp|n with 43,798 marks. OGS used an Intel Core i7 6950X 'Broadwell-E' chip clocked at 5,175MHz (+72.50%). This was joined by a GeForce GTX 1080 Ti with the Pascal GPU clocked at 2,645MHz (+78.72%) and graphics memory at 1,555MHz (+13.01%). His X99 platform motherboard of choice is a GIGABYTE X99-SOC Champion.
In 2D benching to managed to pull off a leading score with a dual-core CPU, hitting a Global First Place score in the HWBOT x265 4K benchmark of 6.6 fps. This is just ahead of Italian xMec with a score of 6.5 fps. The new GFP score was made using a Core i3 7350K pushed to a very nice 6,498MHz (+54.71%). His rig in this case was based on the ASUS ROG Maximus IX Apex board. Using the 1080p preset of the same benchmark he used the same Kaby Lake processor at the same speed to hit a score 28.36 fps, another dual-core Global First Place. In the quad-core rankings he returned to the HWBOT x265 4K benchmark OGS was armed with a Core i7 7700K which he pushed to 6,733MHz (+60.31%) to score 13.29 fps, also a Global First Placed score. With the same settings on the same rig he also took down the 1080p benchmark top spot, hitting a score of 56.18 fps.
OGS is now the No.1 overclocker of Greece and No.4 in the worldwide rankings here on HWBOT with 2,490.8 points. Nice going sir!
A few weeks ago we published an opinion poll here on the HWBOT page (just to the right of this news post). It posed a question related the recent emergence of new High-End Desktop (HEDT) components from both Intel and AMD, asking which which of the available (or soon to be available) processors you would find most tempting this summer. Today, the results are in.
Of course, in recent years the HEDT space has basically been owned by Intel (the company that indeed coined the acronym HEDT in the first place). Intel’s HEDT offerings go way back to the days of Gulftown and its first Extreme Edition lineup. In recent years we've had adapted server chips based on Sandy Bridge, Ivy Bridge, Haswell and Broadwell architectures. These brought more cores, more PCIe lanes and features like quad channel memory to the humble desktop. It also meant getting involved in an eco-system that was much more expensive of course. The Broadwell-E deca-core Core i7 6950X commands prices of around $1,600, making the HEDT space one where only those with the deepest of pockets could play.
Having finally managed to make themselves relevant again in the mainstream CPU space with its Ryzen processor series, AMD plan to launch Threadripper series CPUs that will give HEDT uses a few new options. The Threadripper chip that has so far garnered the most attention from media is the 16 core (32 thread) AMD 1950X, a monster that will offer 64 PCIe lanes, 32MB L3 cache and a ton of grunt. All for around $1,000. Sounds tempting, but according to our poll, apparently not everyone is ready to bite.
In regards to AMD’s new HEDT offerings, more 57% of the 431 HWBOT members polled are keen on shelling out for an AMD TR4 platform, chip with 21.11% of users tempted by the monster TR 1950X. Almost a third of those polled would prefer to stay with tried and trusted Intel platform Core i7 and Core i9 offerings, with the Skylake-X Core i9 7980XE chip being most popular with 10.21% of the votes.
Interestingly however the biggest vote went to neither Intel or AMD. 29.47% of those polled would simply prefer not to buy a HEDT processor this summer. Is this due to prohibitive pricing? VRM-gate? Or perhaps folks are perfectly happy benching with Kaby Lake and Ryzen and just don’t feel compelled to splash this summer? Chime in with your thoughts below.
In his latest PCB breakdown video we find Buildzoid taking a peek at a stripped down graphics card from Zotac, a Macau-based company that is famous for creating smaller components for Mini-ITX and SFF builds. One such product is the Zotac GTX 1080 Ti Mini, a graphics card that attempts to take all the performance you’d expect from a Pascal-based 1080 Ti card, presented in a much smaller design that is gonna fin inside a Mini-IX build. Buildzoid takes a look at the PCB to discover exactly what the Zotac engineers did to reduce the size of the card.
As you would expect, reducing the size of a high-performance component involves making certain compromises. For example he notes how there is a complete lack of input filtering chokes for the pair of 8-pin power inputs. A fact that could become a problem when GPU power draw suddenly causes current spikes, especially when overclocking without a top quality PSU. The vcore VRM features an 8 stage design that uses a uP9511 buck controller from uPI, a pretty much standard controller chip for a high-end Pascal card. The uP9511 is a true 8-phase controller (which means the VRM uses no doubling) with a switching frequency of 600KHz. The VRM is calculated to be capable of 200A at 35W, which as Buildzoid explains is way off what you would see on most hi-end, full length cards.
Buildzoid points out several other key design features that have been made in order to shrink the PCB down to Mini-ITX compatible levels. None of these decisions help make the card real contender for extreme overclocking, which I’m sure is not a huge surprise. You can catch the video from Buildzoid here on the Actually Hardcore Overclocking YouTube channel.
The latest podcast from Hardware Asylum is now available. In Episode 77a Dennis Garcia and Darren McCain take on the topic of digital currency mining and its effects on the PC component market and more. Here are the show notes:
In the time between when this Podcast Extra was recorded and subsequently released we saw Ethereum rise the popularity viral ranks, crash due to fake (or staged) news on 4chan, recover (kinda) and then completely fall out of popularity due to mining difficulty and share price.
It makes you wonder how big corporations actually made money back before the Great Depression and before the advent of stock regulations. it is a little known fact that stock manipulation and deceptive news stories would cause investors to flood or dump a stock based on nothing more than what a respected reporter was publishing. This often lead to quick sales or prompting investors to short a stock in hopes it will go down in price. Stuff like that has become more difficult these days however in the cryptocurrency world something as simple as a fake death story can cause investors to dump everything which will cause the price to plummet as investor confidence declines . It also creates a perfect investment opportunity much like it did for the American stock market some 100 years ago. So, why the history lesson?
Well, I am getting really tired of NVIDIA Pascal based GPUs always being out of stock. At first, during the launch, it was due to demand and low stocks. However, now it is because the superior power efficiency is making GPU mining more affordable. In fact ASUS has released a dedicated mining card based on the GTX 1060 that is just like any other video card except it has no video output. This makes the card worthless on the open market but gives miners a dedicated product for crypto mining with increased reliability while leaving the gaming GPUs to do what they do best in the hands of enthusiasts that know how to use them.
Catch the full podcast here on Hardware Asylum.
In the months following AMD’s highly anticipated Ryzen micro-architecture launch, enthusiasts and overclockers have adopted the new AMD Ryzen processors and propelled AMD back to 2013 activity levels. AMD’s processors stand at a 24.75% share, up from 9.05% in January 2017. Intel’s dropped from 90.84% to 75.17% in the past six months. However during the first half of 2017, Intel’s enthusiast eco-system has shown to be strong enough to withstand the recent surge from AMD as their enthusiast base grew 20% and 16% in 1Q17 and 2Q17 respectively.
The information is based on data provided by HWBOT, an enthusiast and overclocking-oriented organization which tracks benchmark world records and organizes overclocking activities. The data is based on self-reporting enthusiasts sharing their overclocking achievements. It excludes data from applications not compatible with AMD hardware. The data shows a significant uptake in AMD activity from January to June highlighted by a peak jump around 2Q17 time-frame when the AMD Ryzen 7 processors launched globally. It follows similar reports from Passmark AMD vs Intel Market Share charts and LinusTechTips’ 2Q17 Viewer’s Choice PC data.
Despite the positive news, AMD still has a lot of work to do if they want to pose a real threat to Intel’s dominance in the enthusiast space as since 2007, it has not been able to catch 30% of the market.
You can read the full report here on HWinsights
Over the last few days we’ve seen some pretty interesting action regarding the 3DMark03 benchmark with the World Record changing hands a few times. Just last Friday we brought you news that Sweden’s No.1 overclocker Rauf had managed to submit a World Record score in the 3DMark03 benchmark. His record score of 359,451 marks did not last long however. TeamAU just hours later, posted a score of 359,792 marks. In both cases the hardware used consisted of the latest Kaby Lake-X Core i7 7740K processor and an Nvidia GTX 1080 Ti card.
Within hours of the TeamAU submission, Rauf proved that he still had plenty left in the tank, posting a score that broke the 360,000+ barrier for the first time. With a Core i7 7740X chip clocked at 7GHz, a GALAX Geforce GTX 1080 Ti card pushed to 2,550MHz (+72.30%) / 1,600MHz (+16.28%) he managed to hit a score of 360,721 marks, reclaiming the 3DMark03 World Record. Other rig components include an ROG Rampage VI Apex motherboard with DDR4 memory clocked at 2,078MHz (12-11-11-28).
Enter Dancop. The current German and World No.1 ranked overclocker entered the 3DMark03 fray armed with a Titan Xp card with the GPU configured at 2,550MHz (+72.30%). With 3DMark03 being quite an old benchmark however, CPU grunt remains a vital factor. Dancop managed to push his Core i7 7740K to several MHz higher than his competitors, hitting a CPU clock frequency of 7,169.4MHz (+66.73%). The upshot of all this work was a new World Record score of 365,098, well and truly blowing the doors off that 360K barrier!
It’s not every day that we find the regular tech media really willing to pull back the curtain and delve a little deeper into the technology that is driving our modern PCs. Which s why it is pretty refreshing to see Gamers Nexus chief Steve Burke take time to interview two of AMD’s most technically adept employees. A little while back Steve managed to get some quality time with Sam Naffzieger and Micheal Clarke, a Chief Architect from AMD, who is intimately familiar with the processes and strategies that that were used to create the new Ryzen architecture. The whole interview is caught on video, but for those of you who like to read, here’s a sample of their conversation in regards to uOp caching, power optimizations, shadow tags and more:
Michael Clarke:“One of the hardest problems of trying to build a high-frequency x86 processor is that the instructions are a variable length. That means to try to get a lot of them to dispatch in a wide form, it’s a serial process. To do that, generally we’ve had to build deep pipelines, very power hungry to do that. We actually call it an op-cache because it stores [instructions] in a more dense format than the past; what it does is, having seen [the instructions] once, we store them in this op-cache with those boundaries removed. When you find the first one, you find all its neighbors with it. We can actually put them in that cache 8 at a time so we can pull 8 out per cycle, and we can actually cut two stages off that pipeline of trying to figure out the instructions. It gives us that double-whammy of a power savings and a huge performance uplift.”
Sam Naffziger: : “X86 decode, the variable length instructions, are very complex -- requires a ton of logic. I mean, guys make their career doing this sort of thing. So you pump all these x86 instructions in there, burns a lot of power to decode them all, and in our prior designs every time you encounter that code loop you have to go do it again. You have this expensive logic block chunking away. Now we just stuff those micro-ops into the op-cache, all the decoding done, and the hit-rate there is really high [Clarke: up to 90% on a lot of workloads], so that means we’re only doing that heavy-weight decode 10% of the time. It’s a big power saver, which is great. The other thing we did is the write-back L1 Cache. We aren’t consistently pushing the data through to the L2, there are some simplifications if you do that, but we added the complexity of a write-back so now we keep stuff way more local. We’re not moving data around, because that wastes power.
Catch the full interview from Gamers Nexus here on their YouTube channel, plus a written article here which is perhaps more suited to the readers amongst us.