Ivy Bridge-E Benchmarks

[dbcoopernz]

A poster on a chinese forum is claiming to have tested the Core i7-4960X CPU.

http://www.techpowerup.com/183176/C...ghly-10-Faster-than-i7-3970X-Early-Tests.html

PC enthusiast "Toppc" with the Coolaler.com, with access to a Core i7 "Ivy Bridge-E" sample clocked to match specifications of the Core i7-4960X, wasted no time in comparing the chip to a Core i7-3970X "Sandy Bridge-E." The two chips share a common socket LGA2011 design, and run on motherboards with Intel X79 Express chipset. An MSI X79A-GD45 Plus, with V17.1 BIOS was used to run the two chips. Among the tests Toppc put the chip through, are overclocker favorites SuperPi mod 1.6, CPU Mark '99, WPrime 1.63, Cinebench 11.5, 3DMark Vantage (CPU score), and 3DMark 06 (CPU score).

The Ivy Bridge-E chip outperformed its predecessor by roughly 5-10 percent across the board. In Cinebench, the i7-4960X scored 10.94 points in comparison to the i7-3970X' 10.16; SuperPi 32M was crunched by the i7-4960X in 9m 22.6s compared to the 9m 55.4s of the i7-3970X; CPU Mark scores between the two are 561 vs. 533, respectively; 3DMark Vantage CPU score being 38,644 points vs. 35,804, respectively; and 3DMark 06 scores 8,586 points vs. 8,099 points, respectively. In WPrime, the i7-4960X crunched 32M in 4.601s, compared to its predecessor's 5.01s.

 

[ShintaiDK]

SB-E vs IB-E is easy to predict if same cores. Just looke at SB vs IB. There is no surprise.

 

[ViRGE]

SB-E vs IB-E is easy to predict if same cores. Just looke at SB vs IB. There is no surprise.

Indeed. The only real questions are these: are clockspeeds going up, and/or is power consumption going down? And if so by how much?

The architecture itself is well known at this point.

 

[SunRe]

Indeed. The only real questions are these: are clockspeeds going up, and/or is power consumption going down? And if so by how much?

The architecture itself is well known at this point.

I'm almost certain that these parts should consume less power and due to this IvyBridge E should also overclock better. The reason I say this is that the server counterparts are reported to have 10 and 12 cores and I doubt Intel would get past the 150W TDP mark as they have now on the highest rated SB-E, but with only 8 cores.

That should be a pretty good indicator, right?

 

[lagokc]

I'm almost certain that these parts should consume less power and due to this IvyBridge E should also overclock better. The reason I say this is that the server counterparts are reported to have 10 and 12 cores and I doubt Intel would get past the 150W TDP mark as they have now on the highest rated SB-E, but with only 8 cores.

That should be a pretty good indicator, right?

Unfortunately Ivy Bridge doesn't seem to be a better overclocker than Sandy Bridge despite producing less heat. Picky transistors :|

 

[Fx1]

Unfortunately Ivy Bridge doesn't seem to be a better overclocker than Sandy Bridge despite producing less heat. Picky transistors :|

Intels 22nm fab is designed for low power and not high clocks.

Something to do with the substrate used in 22nm.

Intel is abandoning the desktop and the enthusiast in favour selling mobile chips.

This is why Sandybridge was the last decent CPU ever made. Everything thats coming after seems to leave desktop back in 2010

 

[joshhedge]

Intels 22nm fab is designed for low power and not high clocks.

Something to do with the substrate used in 22nm.

Intel is abandoning the desktop and the enthusiast in favour selling mobile chips.

This is why Sandybridge was the last decent CPU ever made. Everything thats coming after seems to leave desktop back in 2010

Evidently not, considering Intel is still making performance gains in the desktop space.

 

[Fx1]

Evidently not, considering Intel is still making performance gains in the desktop space.

So 3-5% are performance gains?

You also realise that IPC applies to both desktop and mobile.

Also i dont see any real gains since Sandy. It looks like treading water.

Desktop enthusiast has pretty much been abandoned.

Intel will still be selling Sandybridge based extreme chips until Q4 this year.

 

[tweakboy]

The key here is, can Ivy E be Oced to 5Ghz @ no higher then 1.400v if all you need is 1.35v for 4.8Ghz.... for a Ivy now.

Im getting the 4930k Ivy E cuz I need more cores for Daw. Their already fast enough per core,,, imagine 12 of them in HT x64 app,,,,,,, youll see more then 10 percent difference,,,,,,, Ill see a 20 percent difference probably. gl

 

[Qianglong]

The i7-980x still looks like one hell of a mighty chip!

 

[tweakboy]

Very nice rig and CPU my friend. Qianglong

I decided to take another route. Get the 3820 until Ivy E 6 core comes out....Then selling the 3820 for 200 dollars or so. Grab the 4930k and slap it on my mobo.

 

[moonbogg]

If this is true, then I hate to say it but why even bother? Lets hope it OCs well. I duno, its not like any game is coming out in the next 5 years that a 3930k can't destroy anyways.

 

[cytg111]

Just as we predicted .. IB-E would be close to irrelevant unless they upped the core count which is why many expected at least an 8-core sample .. guess not.

Why even make it? As a trial/beta run for the true server parts ?

 

[lagokc]

Just as we predicted .. IB-E would be close to irrelevant unless they upped the core count which is why many expected at least an 8-core sample .. guess not.

Why even make it? As a trial/beta run for the true server parts ?

Because the profit margin is enormous - price of a server CPU but support and production costs of a retail CPU.

 

[ShintaiDK]

Just as we predicted .. IB-E would be close to irrelevant unless they upped the core count which is why many expected at least an 8-core sample .. guess not.

Why even make it? As a trial/beta run for the true server parts ?

Dont say you are surprised that LGA2011 "enthutiast" is nothing more than a glorified workstation/server platform and the later being the dominant determinator in products.

 

[cytg111]

Surprised? Yes I am. Whenever I step on something counter intuitive or what seems to be illogical, I begin searching for the hidden variables. All humans should.

I dont get why they would spend the money on certifying IVB-E for 2011 when they allready have a product that does the job. I dont get it.

 

[ShintaiDK]

Surprised? Yes I am. Whenever I step on something counter intuitive or what seems to be illogical, I begin searching for the hidden variables. All humans should.

I dont get why they would spend the money on certifying IVB-E for 2011 when they allready have a product that does the job. I dont get it.

It cost them nothing. The server/workstation segment pays for it. And that segment wants higher performance/watt and more cores.

 

[sm625]

5-10 % is a big floating turd. They could have gotten 5% just by releasing a new stepping of SB-E with 5% higher clocks.

 

[ShintaiDK]

5-10 % is a big floating turd. They could have gotten 5% just by releasing a new stepping of SB-E with 5% higher clocks.

But not the performance/watt gain. Nor the cost reduction. And not the extra cores.

 

[cytg111]

It cost them nothing. The server/workstation segment pays for it. And that segment wants higher performance/watt and more cores.

Well, if it costs them nothing, then I guess the cost/benefit equation plays out.

But really who is the target group here, it is not those who allready have SB-E, it must be those who is in the market for this segment but havent allready gotten the product yet(sb-e).
Maybe 'naive/not feeling the market', but I imagine that is a rather small segment.

 

[cytg111]

..... And not the extra cores.

- What extra cores ?

 

[Lepton87]

How many cores will IVY-EP have? The last hope for enthusiast is the hope that 125MHz and 166MHz straps will work with IVY-EP. I guess at least 10 cores, so 10 cores OCed by 30% would be good. Assuming stock clock of 3.2GHz that would be 4.16GHz. 10 IVY cores at 4.15GHz(4,45GHz in SB terms) vs 6 cores at 4.5-4.7GHz with SB-E. From little loss in ST to almost parity and HUGE advantage in MT. Of course it would cost 2k USD, but people who buy Extreme CPUs instead of 3930K already don't care about money at all so it would be a huge bone to them alas I doubt BCM straps would work on Xeons but it's possible. Maybe it would look even better then I thought at first, 10 core IVE-EP will have more than 15MB of cache, 7% IPC improvement is with the same amount of cache. Maybe more cache would boost that to 12%. 15mb l3 vs 25mb l3 will have a big impact. 8 core Xeon at 3.1 is sometimes faster in ST then the fastest SB-E due to 20MB of cache although that's very rare.

 

[ShintaiDK]

- What extra cores ?

IB (Server) dies comes in 6, 10 and 15 cores. 15 being the EX.

 

[cytg111]

And on the other hand, significantly upping the core count without haswell's tsx .. as a programmer doing parallel stuff, it just wouldnt feel right.

 

[cytg111]

IB (Server) dies comes in 6, 10 and 15 cores. 15 being the EX.

Yea, but we are only talking the E part here, right? .. 6 cores. For a second there I thought i missed something.