Intel Broadwell-E Core i7-6950X Review: The first 10-core enthusiast CPU is a beast - martintagazier1947

Large. Hefty. Or in the parlance of our multiplication: OP for "overpowered."  All are apt terms for Intel's new 10-core Core i7-6950X, a muscular CPU so immoderate in power, that you'd have to exist crazy to need one.

Just Intel's Extreme Version CPUs make never been well-nig catering to practical need. No, they cater to your desire for inexperienced performance.This time, though, Intel is pushing both performance and your wallet to the very edge with a CPU priced at $1,723.

Nobelium, that's non a typo: $1,723. For clean uncomparable CPU.

For much details on the chip, read Mark Hachman's deep dive. To ascertain if Intel's brutishly powerful silicon chip is worth it, read on.

Intel has been pushing the Extreme Edition chips for a decade now just this ane whitethorn top them all.

No surprises Here

Broadwell-E has credibly been matchless of the worst-kept secrets in the industry. Models and spectacles leaks have been mobile around for months. All those rumors are simply that until we tumble from the horse's mouth, and Mr. Ed has finally spoken. Here are full inside information on the Broadwell-E line of merchandise, including the current two Skylake "K" chips (outside the red box.)

The Broadwell-E family includes quatern models all of which cost a little to a higher degree the Haswell-E chips they replace.

The Broadwell-E family essentially replaces the three Haswell-E family chips, which were introduced deuce age ago with the LGA2011-V3 weapons platform. Leave out for the current 10-burden Broadwell-E, which creates a new uber-poker chip tier above the others, the rest fall in line with the older Haswell-E chips they supersede.  See the details below from the original rollout of Haswell-E with the pair of smaller Haswell "K" CPUs (outside the cherry box).

Haswell-E introduced a new LGA-2011 V3 chopine and 8-cores for "free."

Besides the new cores, you also get on a price step-up—a fringe benefit of having basically no competition. Intel continues to declare oneself a"budget" Broadwell-E, which has half a dozen cores and fewer PCIe lanes lendable in the chip. That conclusion isn't technical, it's marketing. If you deficiency to build or buy a PC with 40 PCIe Gen 3 lanes turned happening, you induce to pay the extra price.

An easy upgrade

On the remote, the new Broadwell-E's heat spreader gets a Thomas More angule design that increases its durability for the more delicate 14nm chip inside. Corresponding Haswell-E, it uses solder interface textile rather than thermal glue.

Broadwell-E was e'er intentional as a drop-in successor for Haswell-E thus for the most part (more connected this by and by) there's no surprises. Just update your BIOS and socket in the chip and you're ready to rocket. The red-hot buffalo chip also supports DDR4/2400 officially, which Haswell-E did not (though it did just fine thereupon memory anyway).

Intel's new 10-core Broadwell-E sits midmost, with an eighter from Decatur-core Haswell-E connected the left and an older Haswell quad-Congress of Racial Equality on the right.

Broadwell underneath

The real microarchitecture exclusive shouldn't storm, either. Information technology's built happening a 14nm process using the Broadwell (5th-gen) cores that throw been in laptops since 2022 (late 2022, if you count Core M). Broadwell in reality made a very late appearance in desktops in the unwanted Core i7-5775C (which I reviewed) in 2022 before quickly sinking into obscurity when the 6th-gen Skylake CPUs showed up days after.

Hither's a colorful of the die. As you can see, it's a native 10-core chip on the highest-end Broadwell-E part. Intel doesn't pull any funny business by victimisation a chip with 12 cores and turning off two. The 8-gist and two 6-core models use the same 10-substance chip, with cores permanently switched off.

The Broadwell-E give-up the ghost is a native 10-core part with all of them horny. That mystery blacked out portion isn't Area 51. Intel says IT's Xeon functionality that isn't used.

Turbo Boost Max Technology 3.0

Among the almost notable changes to Broadwell-E is the new Turbo Boost Goop Engineering 3.0 feature.

Turbo Rise was introduced with the forward Core i7 chips in 2008. Like the name says, it temporarily increases the clock speed of the chip to improve performance.

The Turbo Boost Max Technology 3.0 substitute lets you set an app to keep going the fastest essence.

Turbo Encourage Max 3.0, exclusive to the Broadwell-E, is quite antithetic. Intel said it identifies at the factory which CPU CORE is the "top-grade" and runs IT at a higher time speed than the others.

Turbo Boost Easy lay 3.0 can then hold fast single-threaded applications to that one higher-flying core, for a performance supercharge of up to 15 percent.

Turbo Boost Max 3.0 can cost increase apps in the foreground, and information technology lets you set apart a uncommon app to a particular core Beaver State cores.

 In Windows, you've been able to bind a fated computer program or process to a particular essence operating theatre wander aside changing the affinity. Turbo Boost Max 3.0 does IT for you automatically (once set upwardly).

The BIOS of the Asus X99 Deluxe Two control board shows the list of the cores and the asterisk indicates what is the "uncomparable" CORE of the bedding material.

Per-core overclocking and more than

Broadwell-E has a couple of features catering specifically to the overclocking sports—nerds who force CPU time speeds to multi-gigacycle levels using liquid nitrogen and other unusual cooling methods in competition. One feature, for instance, lets you crank back the AVX ratio to lour the power it consumes during benchmark runs.

Not everyone is into extreme overclocking sports, though. Many just want to tune a CPU to its highest actual usable performance. For them, Intel has added per-core overclocking.  With elderly CPUs, overclocks were somewhat granular in this you could pick higher frequencies based happening whether it was exploitation, say, two cores. The cores picked, though, were random. With Broadwell-E you can overclock a particular core and even exchange its man-to-man voltage.

When combined with Turbo Boost Max 3.0, you could, say, set an application to run on that particular overclocked core . That pays real dividends in performance for someone willing to put in the tuning time.

Lots of cores, lower clock speeds

It's no coincidence that so many of Broadwell-E's innovations are aimed at giving the chip better performance in single-threaded or lightly-threaded tasks. That's because Intel knows applications that force out take chockablock vantage of the resources of a 10-core chip, or even an 8-core chip, are rare.

That awkward truth has e'er put the company's big chips at a distinct disadvantage to the littler quad-cores so much arsenic Intel's Core i7-6700K. With fewer cores and lower thermal command overhead, those quadrangle-cores commode easy run at higher time speeds. For instance, the same Core group i7-6700K has a base clock rush of 4GHz, while the top-closing 10-meat Broadwell-E has a base clock of 3GHz. All the tweaks Intel has put into Broadwell-E, the keep company said, should redact it on far better footing with a nimbler quad-core, patc liberal IT the capacity to blow the doors bump off when the load needs more four cores.

Read on for our performance benchmarks you bet we tested

Performance and how we tested

Gordon Mah Ung

Here's the backside of Intel's new 10-core Core i7-6950X chip.

For our performance examination I set up three different systems to test the 10-core, nonnegative three chips that I think should be compared to IT: An eight-core Haswell-E Core i7-5960X, a quad-core Skylake Essence i7-6700K, and a six-core English ivy Span-E Core i7-4960X.

All three systems received clean installs of Windows 10, and each was tested with a GeForce GTX 980 lineup and twinned Kingston HyperX Violent SATA SSD. All three also had 16GB of Barbary pirate RAM. We put-upon DDR3/2133 in triple-channel mode for the Hedera helix Bridge-E system, and DDR4/2133 in quadruplet-channel mode for the Haswell-E and Broadwell-E systems. The Core i7-6700K ran in dual-channel mode.

The Haswell-E and Broadwell-E were swapped into the same organization for testing. The fashionable available UEFI builds were also installed on all three motherboards. Both the Skylake and the Ivy Bridge-E chips used Asus motherboards, while an X99-based Asrock board was used for the Haswell-E and Broadwell-E CPUs.

Unmatchable bear-sized caveat

Earlier we get too far into the benchmark-o-Rama, I'd like to signalise that I ran into a snag archeozoic on that simply could not atomic number 4 remedied in time to make this tale deadline. Intel has maintained that Broadwell-E is completely drop-in compatible with existing X99 motherboards on the food market. That manifestly means it'll mould, simply it doesn't skilled all of the features exploit. When I tried to install the required number one wood and service program for the Turbo Hike up Max 3.0, it bombed out because the Asrock X99 Extreme4 board doesn't support information technology.

When it'll be added I don't know. Intel said financial support can be easily added through and through a UEFI update, but IT's up to the individual board vendor to do so. In other speech, the Numbers you catch here for single-rib tasks, which could be up to 15 percent faster in theory with Turbo Boost Max 3.0.

Very late in the work on, I was healthy to get the chip into an Asus X99 Deluxe II board. That solved most of my problems, but I didn't have time to re-outpouring all of my tests. The complete news is the Turbo Boost Max 3.0 should impact only the single-threaded tests.

CineBench R15 performance

In multi-threaded tests, those 10 cores pay very nice dividends.

We'll start this away with a test that's ideally clad for a 10-core chip: Maxon's CineBench R15. This is a benchmark supported Maxon's Cinema4D rendering engine, which the company uses in its transaction products, so you can consider it a reflection of real-world performance.

CineBench R15 loves CPU cores, and the result is pretty ugly. The Broadwell-E blows past the 8-core Haswell-E chip and stomps the quad-core Skylake chip. You have to give proper quotation to it Skylake chip, though: Combined with its progressive 6th-gen cores and its 4.2GHz clock hurrying, IT really punches preceding its sort out.

To stupefy a wee much insight into how the various CPU cores do when you Don River't factor out the difference in core count, I also ran CineBench R15 in the optional single-rib mode. The high time speeds summation the newer 6th-gen cores put away the Skylake chip in the foremost seat by a very healthy edge. The Ivy Bridge-E chip does fairly well, merely running at a higher clock pep pill, likewise.

The worst score comes from the Haswell-E, which I'm going to impute to its lower clock speeds. Broadwell-E does particularly fit at 3.5GHz and this is without Turbo Boost Max 3.0 on.

CineBench R15 in unmated-rib performance gives the nimbler, higher clocked Skylake chip the edge.

Blender performance

The second benchmark I'm going to detail is Blender, a  free and popular 3D renderer used for visual personal effects by many indie flic makers. The test file away I used was Mike Pan's free BMW bench mark file.

The 10-core Broadwell-E still leads the pack, but past less than we expected. I've also seen Blender not offer the same core scaling as I've seen unsuccessful of Maxon's Movie house 4D locomotive. Going from a dual-core to a quad-core laptop has also shown hardly average scaling.

The upshot is if you'ray working on your indie film task and all the put to work is done in Blender, you'd cost satisfactory with a quad or six-core part. But hey, if you're an indie filmmaker in any case, you should be working connected a shoestring budget, not dropping $1,723 for a Central processor.

The popular and free Liquidizer render engine likes multi-core but not quite as often as Maxon's engine.

POV-Ray Performance

I'll close outer my 3D rendering test subdivision with POV-Ray. This 3D graphics program dates back to the Amiga and is available gratis. We see very nice scaling from the 10-core Broadwell-E. Probably enough to warrant the expense if you really are doing POV-Ray projects and your renders are teeth-gnashingly long.

And yet again, that six-core Common ivy Bridge deck-E Core i7-4960X is starting to look pretty mouldy against the quad-core Skylake Core i7-6700K chip.

Nobelium surprise, POV-Ray also likes 10-core CPUs.

DirectX 11 gambling performance

Get me just get this out of the direction past saying that no, in today's gaming experience, a 10-core Central processing unit doesn't get you to a greater extent performance. Information technology just doesn't. That's because the vast majority of games don't exploit completely those cores. Even the highly touted DX12 in all likelihood South Korean won't party whip games into shape for at least some other year OR deuce. Still, you want cogent evidence so the first matter I'm going to run is the DirectX 11-based Tomb Raider.

Again, all of our tests were keep going a GeForce GTX 980 with the Sami driver. For my runs, I set out Tomb Raider at 1920×1080 resolution victimisation the normal preset. Mostly, IT's a tie. The real storm is how Haswell-E and Broadwell-E pull ahead by a spot. Even the Ivy Bridge-E is technically faster, simply let's non kid ourselves. This is a linkup. I could run another six more games, but all you'd see is a tie across the vast majority of games. Gaming is shut up mostly 80 percent well-nig the GPU.

The lesson here is if your system of rules is primarily utilised to play one back at a time, you don't need more than a quad-core micro chip with Hyper-Threading.

No, you father't baffle more performance out of a 10-heart chip in gaming but you already knew that correctly?

DirectX 12 gaming carrying into action

Yes, but there's that DirectX 12 thing right? You live, the big move from Microsoft to gain its play API actually exploit the three-fold CPU cores we've had in our PCs for the half a decennium.

To examination it I first ran 3DMark's DirectX 12 feature test. It tests a PC's ability to issue draw calls or suck in objects to a screen. You bathroom see the Skylake Core i7-6700K gasses out at the 5-million-absorb call number. We then see a huge bump to the Common ivy Bridge-E chip, and and then we fundamentally flatline from 12-threads all the way to 20-threads.

The upshot from the 3DMark DX12 have test is you don't seem to actually need more than than a  6-core group CPU with Hyper-Threading.

You seat picture 3DMark's DirectX 12 feature exam can't punctuate more than the 12-threads of our English ivy Bridge over-E chip.

Ashes of the Singularity performance

But what about a real game? To find out I stone-broke out Oxide's Ashes of the Singularity, a new proper-meter strategy game that is the showcase title for DirectX 12 performance and absorb yell capableness. Eve break, Oxide provided us with a beta interlingual rendition of Ashes of the Singularity that adds a new manner specifically to quiz CPU performance, rather than GPU performance.

The scene adds a larger map and more complexness using the duplicate engine to push CPUs harder. Oxide said it's withal in the process of tuning the bench mark but was willing to rent us run information technology early.

The result is certainly a picayune more promising for the new 10-core. In the return-on-investment category, however, at least at this peak in Ashes, information technology has yet to justify a $1,723 spending. We'll plan to revisit this test when it's finalized.

We got former access to a new CPU focused test in Ashes of the Singularity pun to test multi-core carrying into action subordinate DirectX 12.

3DMark FireStrike Physics

We'll close gaming performance with a score from the physics portion of 3DMark FireStrike. It simulates adequate 32 threads of game physical science using the Fastball Open Source Physical science Library that's too used in such popular games as Wonderful Thievery Auto V and Red Bloodless Redemption .

Here we see a pretty hefty advantage for the 10-pith chip. The surprise is the gap it opens between the 8-core Haswell-E chip.

While information technology's a triumph for the Broadwell-E, I make to point extinct that this is a theoretical win, as a couple of game developers are adding enough game physics to actually take 20 threads of computing. If they ever did though, that 10-core would atomic number 4 king.

The physics test in 3DMark FireStrike gives the 10-core a very nice bump.

WinRar compression performance

To measure how fast the 10-heart knap pushes densification, I used the built-in benchmark in RARLab's WinRar. WinRar loves gobs of togs, and the 10-core Broadwell-E again opens awake a can of whup-seat on all others.

The results came as a surprise to Pine Tree State, simply if your Clarence Day chore is compressing files in WinRar, a 10-core might be worth it.

If you behave lots of compression, you'll want scads of cores it looks like.

7Zip compression performance

Interestingly, information technology's not just WinRar that loves multi-threading. I also laid-off up the freeborn and superpopular 7Zip compression utility to get a line whether the WinRar results were fooling my eyes. 7Zip's bench mark lets you choose the workload based on the maximum amount of duds in the system. For each CPU, I duplicate the workload to the togs each chip has, so 20 for the Broadwell-E and 8 for the Core i7-6700K. The results, over again, put the 10-core well onward of all others. Nicely done, Broadwell-E.

7Zip loves dem cores too!

HandBrake 10.2 performance

For our encoding test, we take in a 30GB 1080p MKV file and transcode it using the slaveless and superpopular HandBrake utility. Our target file format and size uses the Android tablet preset. The results Here put the 10-core Broadwell-E in front, but I'm actually disappointed a tad. Sure, you knock off off a good ball of time in an cipher, but that Nitty-gritty i7-6700K is close behind.

Handbrake 10.2 likes cores but our workload doesn't seem to tenseness it as some.

Premiere Professional Creative Cloud performance

For a TV exam, I used Adobe brick's First In favour of Imaginative Cloud picture editor. Information technology's a hugely popular video editor with professionals and prosumers. Premiere Pro supports both GPU and CPU encoding, but to find out which CPU was the fastest, I opted for CPU encoding. The workload was a 4K project Intel provided. I tried using an actual working project created by our own TV team .but I found the 1080p video from our Canon C100s didn't push our CPUs hard enough, with all four chips finishing our encode nearly concurrently.

Intel's test files increase the resolution to Ultra HD 4K and are a bit more work. The result though, isn't all that impressive. The Broadwell-E is the fastest, only that quad-core Skylake does pretty well considering its nub count. Oddly, the 8-core Haswell-E underperformed too—I'm not sure why, just multiple runs all produced the same solvent. Single theory is the Ivy Bridge-E chip seat run up to 4GHz, while the stock Haswell-E clocks in around the low 3GHz range. Mayhap this workload favors the higher clock speeds of the Common ivy Bridge-E chip and really doesn't need more than 12 threads to run. The whole win still goes to the Broadwell-E by a healthy amount, but Iexpected more.

caption goes here

HWBOT x265 execution

For my 2nd encryption test, I decided to throw HWBot x265 at my test CPUs. Information technology's a test created by Czech overclocker Havli and is built around an ASCII text file x265 encoder. It's a punishing run. It loves CPU cores and supports numerous  modern advanced instruction sets such as AVX2 and FMA3.

The 10-core wins this again quite handily. The interesting side note is yet though the English ivy Bridge deck-E has more two more cores than the Skylake chip, the newer instruction sets and efficiency of the 6th-gen chip appear to give it a squeamish edge.

How good is that final result? Non bad. The world record at overclocking enthusiast site HWBot.org is held by Slinky PC, who hit 12.59 fps using a 22-core Xeon E5 2696 V4 break away, apparently overclocked to 3GHz.

Caption goes here

Megatasking performance

Even though the new 10-inwardness Broadwell-E is a monstrous chip in many multi-rib apps, you Crataegus oxycantha be disappointed that it doesn't just yaup on the musculus quadriceps femoris-core Skylake chip by huge margins. It does, after totally, have six many cores at heart.

For one, IT has a lower stock clock speed. Overclocking the Broadwell-E gets both within spitting distance; but on multi-threading, that Skylake silicon chip bequeath hang in there in just about apps that just rump't use all the cores on the 10-core Broadwell-E.

So what happens if you throw multiple tasks at it at the same time? To ascertain out I fired astir Premiere Pro CC over again and began rendering out a one-hr 1080p television. I then went ahead and ran CineBench R15 on completely iv chips.

The 10-meat Broadwell-E is majuscule—if you bump to like to come quadruplex, multiple compute heavy tasks simultaneously.

The result is probably more succeeding with the beat-shoot down you would expect from a 10-marrow chip compared to that space-essence. While interpreting on the quadrangle-core Skylake, the CPU was squirting at most 100 percent capacity. The 10-core Broadwell-E under the same laden was cruising along at 55 pct. The quad-sum Skylake chip, in fact, was indeed slow that I was able to draw CineBench R15 tercet times on the Ivy Bridge-E before the Skylake chip finished running it once. And then take that, Skylake!

The result is you dismiss run a 4K Premiere Pro CC encode while linear another one or cardinal content innovation apps without seeing everything mire, care it would with a quad-core.

Is this realistic? For a mental object foundation individual or what Intel calls a "megatasker," yes. Most people start onerous compute tasks and take a walk around the pulley-block while it finishes. With the 10-heart and soul Broadwell-E you could maintain working. If time is money, the 10-core chip is the uncolored choice.

It's hard to occupy all the threads available on a 10-core potato chip unless you megatask information technology.

Side by side awake: Conclusion and Overclocking

Intel

How does IT overclock?

I'm always a little hesitant to issue proclamations of how a new CPU overclocks supported a sample of peerless. Galore times information technology's not most the overclocking capableness of the chip, it's about the overclocking capability of the overclocker.

Intel itself, as usual, won't say anything roughly what to have a bun in the oven. That's understandable, as overclocking is normally a crap dash. I can at least come to my personal experiences.

World-class, I'd suppose it depends on the motherboard. The Asrock X99 Extreme 4 on which I ran most of my commonplace time tests was a dismal fail. I couldn't force out the chip much beyond hackneyed and gave up afterward wasting an hour hard to get minimal overclocks out of the chip.

Very late in my retrospect though, I accepted an Asus X99 Deluxe II board. A newer motherboard that supports Turbo Advance Goop Technology 3.0.

With the Asus X99 Deluxe Deuce board, I dialed up a 4GHz all-core, ratio-based overclock and booted into the OS. No muss, no fuss. That's without having to mess with electric potential either.

That's really not bad, and I'll constitute the number 1 to say I am not even remotely an extreme overclocker. I didn't do a formal stability test, but I was able to run numerous multi-threaded benchmarks without event for several hours. I and then overclocked the "best" core up to 4.5GHz and used Turbo Further Max 3.0 Technology to bind particular applications to it.

Overall, I'm pretty happy my sampling overclocked on the Asus X99 Deluxe II. Non so on the Asrock.

But what should you expect? It's still very early in the life of this chip. After speaking to various motherboard and system vendors, it sounds like you should expect at most 4.2GHz to 4.3GHz overclocks happening all 10 cores. On the far side that, I'm told, it gets difficult to manage the heat and emf. Its predecessor, Haswell-E, generally ran out of gas at 4.5GHz in practical use, so Broadwell-E seems to embody within expectations.

You May bounce at the intellection of loosing a infinitesimal general overclocking head room, but the greater efficiency of Broadwell cores over Haswell cores make water up for it.

Broadwell-E cores running play at 4.5GHz pull flat with Skylake cores running at 4.2GHz.

Conclusion

First, I'll tot the performance aspects of the 10-core Broadwell-E past saying, damn, it's a freaking monster. In multi-threaded tasks IT easily thrashes the 8-core Haswell-E. Hyphenated with per-core overclocking and Turbo Boost Max 3.0, information technology can hang with the nimbler Core i-6700K chip in lightly threaded and single-rib tasks, too.

That's a win regardless how you cut it. Intel said it aimed to consecrate you the superior of both worlds for multi-rib and lightly rib, and IT has achieved that.

The elephant in the room is that $1,723 price tag.

Initial rumors last year indicated the 10-core check would slot in at the same $1,000 price of the 8-core Haswell-E. A grand may seem excessive but if you got the 10-core version for the price Intel used to charge for an 8-core, IT's like getting "free" stuff.

Intel actually did just that when it replaced the $1,000 6-core Core i7-4960X with the $1,000 8-core Core i7-5960X chip. Intel isn't handsome away any freebies this time though.

At the price Intel wants, you could actually buy a 14-essence Xeon. That Xeon, though, would belik be even more overkill and would non consecrate you the single-rib performance of the Nitty-gritty i7-6950X.

As it stands, the Core i7-6950X is easily Intel's most mighty consumer break off that it's ever made by a long shot. I just wish it were actually affordable.

Intel

Source: https://www.pcworld.com/article/414986/intel-broadwell-e-core-i7-6950x-review-the-first-10-core-enthusiast-cpu-is-a-monster.html

Posted by: martintagazier1947.blogspot.com

Share this post