Today is the opening day of Computex 2016, and to mark the occasion Intel has a brand new processor family. Designed for the hardest of hardcore enthusiasts, Intel’s new 14-nanometre Broadwell-E chips, the Core i7-69XX and i7-68XX, are its most powerful ever. Forget your garden-variety quad-cores and dual-cores; the Intel Core i7-6950X is an entirely unlocked, overclocking-friendly 10-core monster with support for quad-channel RAM and four graphics cards. If you can’t afford the circa-$2200 price tag for the newest top-end silicon, though, new 8- and 6-core CPUs are also on the way.
This post originally appeared on Gizmodo.
[clear]
[clear]
The new i7 Extreme Edition CPUs follow a long tradition — a decade, in fact — of top-of-the-line, bleeding-edge chips from Intel. The Extreme Edition line started in 2003 with a single-core Pentium 4 Extreme Edition (newly with Hyper-Threading) and since then, every couple of years the number of cores has jumped. The first i7 EE was the i7-3930K with six cores and a 3.2GHz base frequency, but the new i7-6950X takes things to an entirely different level.
In terms of pure numbers-on-paper performance, the i7-6950X and its lesser i7-6900K, i7-6850K and i7-6800K compatriots are performance monsters. With a 3.0GHz base clock the top i7-6950X is actually slower in raw Hertz than its lesser variants, but wins out with more shared cache and a larger number of physical cores and virtual threads supported simultaneously. If you’re looking to upgrade, though, that $US1569 price point might be a little unrealistic — it’s almost $2200 in direct Australian currency after conversion, before shipping and Australia Tax — and the $US587 ($817) and $US412 ($574) 6850K and 6800K look a lot more reasonable.
Intel is saying the new chips can fulfill the same task that Twitch streamers and Let’s Play YouTubers have previously been using two PCs for. Instead of using one PC for gaming and another to transcode and upload the digitally captured footage from that machine, a 10-core i7-6950X CPU can handle all of that simultaneously while operating at peak performance and not dropping frames during gameplay. And the new chips support up to 4 discrete graphics cards, as well as Intel’s fast I/O with Thunderbolt 3.
Here’s a table with the new Broadwell-E processor family’s core specs, and a comparison to Intel’s Extreme Edition chips of old, as well as the chipsets that the EE processors have been running on since 2013. The motherboard side of things hasn’t evolved hugely since the X79 chipset for Ivy Bridge-E, apart from support for faster 2400MHz DDR4, but the processors’ shared cache, the number of physical cores available and Turbo Boost Max 3.0 processor clock adjustment is where the new CPUs set themselves apart.
[clear]
[clear]
Turbo Boost Max 3.0 is an evolution of Turbo Boost 2.0 in the same way that Nvidia’s GPU Boost 3.0 evolved on its ancestor. Turbo Boost Max 3.0 improves the new Extreme Edition CPUs’ single-threaded performance by benchmarking and marking the fastest possible core on the entire die, then assigning critical single-threaded workloads to that most capable core through Intel’s drivers. Intel’s technical team was at pains to tell us that “this is not overclocking, this is a sustained thing,” designed to make the EE chips competitive with faster-clocking quad- or dual-core Skylake i7s or i5 despite lower frequency speeds per individual core.
Intel claims 35 per cent faster multi-threaded performance versus the previous generation’s top chips. In its presentation, Intel’s tech team talked about “megatasking” — a new buzzword they’ve come up with that refers to more complex workloads than multi-tasking. When you multi-task, Intel says, you might have half a dozen different Chrome windows and Word documents open, unrelated tasks. When you megatask, you’ll be mixing and matching similar processor workloads and disciplines. Here’s an example they described: “On these platforms we ran Rocket League in 4K, we made sure that while it was running it maintained 60 frames per second — a 4K game at 60fps. While that was happening we streamed it on Twitch at 1080p. With those two things happening in parallel, we captured the gameplay for around 9 minutes and transcoded it from 4K to 1080p, then we went to upload it to YouTube.
“This is what gamers typically do. And the transcoding piece ran 25% faster on Broadwell-E compared to Haswell-E: you can do this ‘mega-tasking’ scenario and see better performance versus the previous CPU.” When you have applications that work together, like video editing plus animation rendering and post-processing effects or roto-scoping, Intel is confident its new chips will outperform old ones and any competitor’s product. The new i7 Extreme Editions’ on-chip specs — 40 PCI-Express lanes, up to 25MB of cache, may only be a small upgrade from the previous generation, but it’s the behind-the-scenes stuff that matters most.
The new chips support overclocking and are fully unlocked, and all have the same heft 140 Watt TDP running on the existing LGA-2011v3 socket. Overclockers will be happy to hear about new single-core overclocking features, with independent clockspeeds possible on a single chip: “We now allow individual ratios on a per-core basis, so you can set individual frequencies per core.” Tried and tested combinations of motherboard, RAM and CPU will also be certified for certain levels of stock and overclocked performance: “There’s an effort where memory vendors and motherboard vendors will work together to test overclocked motherboard/BIOS/memory setups to confirm their ability — and we will post that on our XMP overclocking website.”
We don’t have confirmed release dates for Intel’s new Broadwell-E chips, nor confirmed Australian pricing, but we’ll let you know as soon as we get it. Don’t expect them to be cheap!
[clear]
[clear]
Comments
26 responses to “Intel’s New 10-Core Extreme Edition Broadwell-E CPUs Are The Most Powerful Ever”
There’s been one for most I7 generations – Nehalem had the I7 965X and 975X, Westmere had the 980X and 990X.
980X is still up to the job. Not bad for a 6 year old CPU. Graphics card is my bottleneck, not CPU (not with a multithreaded app, anyway…).
I was going to say that too I had a 980x it’s a cupboard ornament now. Like you with a old GPU 580OC
*Looks at his AMD rig and sighs*
Eh. If your PC does what you need it to, you don’t need to buy into the ‘shiny new tech’ hype. Now excuse me while I go fist-fight 200 people for a new GTX 1080… *ahem*
Initially read that as “…excuse me while I go fist 200 people…”
Sacrifices must be made.
Was very happy with my AMD rigs. Cost me less for similar performance.
And my feet were nice and toasty in winter. 😉
Kinda disappointed. The X’s have long been a $1k chip, which is perfectly fine IMO. This is starting to become straight gouging.
The 6950X is a new tier is why. Broadwell-E is running four models instead of the three the previous E generations have run. The 6900K appears to be the direct successor to the 5960X and is priced at roughly the same point (if rumours are to be believed).
We also got core count increases each gen with the X’s staying at the same point though. I understand what your getting at, but come on.
That part has been a little less predictable. Can’t really do tables here but the generations and their cores have looked like this (newest generation first):
Broadwell-E: 10/8/6/6
Haswell-E: 8/6/6
Ivy Bridge-E: 6/6/4
Sandy Bridge-E: 6/6/4
Gulftown: 6/6/6
Bloomfield: All 4 cores
It seems that for several generations, 6 was it. Haswell was the first to introduce an 8-core CPU so it depends how you look at it. The trend would suggest they’d sit on 8 for a while so a 10 core is a little unexpected in that sense. On the other hand, you might expect they’d go to 8/8/6 (ie. bring 8-core to the second tier CPU) but they haven’t done that.
I mean, this CPU is definitely not for me. It’s expensive and until games support MT properly it’s an inferior choice to the 6850K with its much higher clock speed. I just feel like I can see where they’re coming from with that entry and it doesn’t seem quite as bad in context.
I own a 5960X, and can always use more grunt for what I do :P. So probably more me being butthurt tbh. Still. :\
Fair enough =) I agree it’s a bit of a shock at first. I do a fair bit of intensive processor work myself (software developer) so I tend towards multicores myself, but I found a while ago that the price-to-benefit ratio beyond your typical i7 CPUs (like the 4770K) starts to become unjustifiable, for my budget at least anyway.
I can’t speak in-depth on the manufacturing process but without knowing in detail how all the materials are gathered and research goes into I can fairly say it is price gouging when margins may be what they are :\
I don’t care about PC building and CPUs, but I do care about graphic design, and that box design is http://3.bp.blogspot.com/-BwpRmOfmpXM/UKke681QoLI/AAAAAAAAANo/JvkG081BFt0/s1600/muah.jpg
I’m still rocking the i7-3930K from 2011 I don’t think ill be upgrading any time soon still a good chip a good overclocked as well
inb4 “my 2600k is still good”
oh dear too late
It’s pretty mental. The high-end chips and components have had longevity not seen before in PCs.
And yes, your 2600K is still good… 🙂
http://www.tomshardware.com/reviews/cpu-hierarchy,4312.html
I have had 2600k and its nothing compared to my 5960x
Okay, me and my 2500k will see ourselves out…
Can anybody who understands well this stuff tell me why Intel breezed through i3 and particularly through i5, but i7 has been around forever?
They’re still making I3 and I5 as well as I7. It’s really a badge denoting relative position in the performance scale, kind of like car models. GSi vs GT vs GTi, for example.
I3 – dual core, no hyperthreading.
I5 – quad core, no hyperthreading.
I7 – quad or more cores, hyperthreading.
On top of this, there’s the generation of the chip – the first Core I3/5/7 core was Nehalem, which replaced the old Core/Core 2 cores which had names like Merom and Allendale.
I7 is the performance beast, and so it seems like it’s around longer because it’s first off the block each generation, followed by the I5s and then I3 budget chips. The first I7 was the 920, 940 and 965X release of 2008. 45nm die. Each year or two there’s either an architecture change (a new core model) or a die shrink (making the previous architecture smaller and more efficient/powerful).
Did that help?
Yes, thanks. So what would you say is generally better, a just-released i5 or an i7 for a previous generation that now costs around the same as the newest i5?
For what kind of workload? Gaming, 3D rendering, compiling, mathematical simulations, day to day computing?
For gaming you should be looking almost entirely at clock speed, whichever is higher is better.
As ZombieJesus says, clockspeed…
My view is it depends on the rest of your rig too. If you’ve got a good mobo, lots of RAM and the CPU is a bit low end, then keeping the rest together and getting the best CPU you can might make sense. If it’s all a bit budget, then you might take a different path and build a new PC.
I’m running a Gigabyte X58-Extreme motherboard. It was high-end, but I put an I7 920 in it, knowing I’d upgrade that later to an extreme in the next part of the cycle.
i3 and i5 CPUs are still around and going strong. They have (or had?) two new Skylake i5 models due just this month and there were some Skylake i3’s about 6 months ago from memory.
The i3 doesn’t tend to get much coverage because it’s the budget range. You’re more likely to find them in budget Dell PCs than gaming desktops these days. The i5 though is quite strong still in gaming because it offers high clock speeds for cheaper than their i7 counterparts.