Intel Merrifield SoC benchmark scores online

[CHADBOGA]

The wonderful Elmer Phud brought this to my attention. :awe:

Intel Merrifield SoC benchmark scores online

Interesting, some benchmark score of Intel's Redhookbay, which is equipped with a dual-core Merrifield SoC (for smartphones), hit the web. In the graphics performance NenaMark2 benchmark the chip reaches a score of 58.30 fps (but that might be limited due to a 60 FPS cap).

In the more general AnTuTu benchmark it scores 31612 points (!) and that is nearly three times as much as the Exynos 4412 based Samsung Galaxy S III and the Tegra 3 based HTC One X.​

 

[sm625]

A wide variety of hardware scores in the 10000 point range. 30k or even 50k scores would be impressive. It would be like intel releasing an atom that pulls a 10 second super pi. Either that or antutu is just a broken benchmark. Given how awful their rankings are organized on their site, I tend to believe the latter.

 

[ShintaiDK]

Those benches are not Merrifield. But Z2580.

 

[Nemesis 1]

Pretty fast phone there but I not interested in a phone until 22nm intel cores. that will be 60 + % faster and with double the battery life

 

[Nemesis 1]

Those benches are not Merrifield. But Z2580.

Link. I haven't a clue . But like you I assumed that was a tablet core

 

[ShintaiDK]

Link. I haven't a clue . But like you I assumed that was a tablet core

 

[Nemesis 1]

A small google and Ya it looks like its the real deal .

http://www.intomobile.com/2012/05/14/intel-merrifield-system-chip-may-end-up-2013-superphones/

So this is the Silvermont core . So look at that hardly any info than this . Now I need Motorolas version and I am a buyer.

 

[Arzachel]

A 1.5Ghz dual core A15 seems to score ~14k, making the score decent to meh depending on how well the benchmark scales with multiple cores.

 

[Nemesis 1]

Ya so the supposed bench shows the single core silvermont will be as fast as the 2 core A15. The dual core Silvermont will equal a 4 core Arm A15 and the 4 core intel will have to deal with AMDs junk. All this at lower power levels than Arm can reach . I will bet anyone that the 4 core Silvermont will not have a 4 eu graphics processor . As Arm is not the comp. in the 4 core small core arena . Its AMD but not in tablets . Amd needs to bring a 7 watt chip to the game that BEATS Silvermont.

 

[Exophase]

Ya so the supposed bench shows the single core silvermont will be as fast as the 2 core A15. The dual core Silvermont will equal a 4 core Arm A15 and the 4 core intel will have to deal with AMDs junk. All this at lower power levels than Arm can reach . I will bet anyone that the 4 core Silvermont will not have a 4 eu graphics processor . As Arm is not the comp. in the 4 core small core arena . Its AMD but not in tablets . Amd needs to bring a 7 watt chip to the game that BEATS Silvermont.

1) This is not Silvermont. 2GHz 32nm Saltwell cores. Of course I'm skeptical that real phones will allow you to run two cores at 2GHz, particularly for any length of time, when Clover Trail only lets you run 1.8GHz in tablets...
2) CPU performance is only a fraction of the total score. GPU performance plays another huge role, among several other categories. Read the website.
3) This benchmark is embarrassingly threaded (probably runs the same instance multiple times with zero interaction), so all the best scoring devices are quad core Cortex-A9s instead of dual core Kraits or even Cortex-A15s. But in most real world stuff the story is very different. Note that while Z2580 is dual core Atoms still get exaggerated bonuses on
4) 1.8GHz Exynos 4 (Galaxy Note II) scores 27598 in Antutu: http://antutulabs.com/Ranking Considering Z2580 will have SGX544MP2 probably at a pretty high clock speed it should easily have more GPU power than the 400MHz Mali-400MP4 in Exynos 4.

All of these things considered the benchmark doesn't really tell us that much, assuming it's real. At the very least we'd need to see the full breakdown of the score. Concluding from this that one one Atom = two Cortex-A15 cores is just.. I don't even really know what to call that.

 

[sm625]

What is the difference between a galaxy GT-N7100 and a GT-N7108? Because one scores 20% higher than the other. So if we can quantify that difference we may be able to quantify just how powerful a chip it would take to score 50k. Can anyone make a guesstimate of what an iPad4 would score if it could run this?

 

[Nemesis 1]

1) This is not Silvermont. 2GHz 32nm Saltwell cores. Of course I'm skeptical that real phones will allow you to run two cores at 2GHz, particularly for any length of time, when Clover Trail only lets you run 1.8GHz in tablets...
2) CPU performance is only a fraction of the total score. GPU performance plays another huge role, among several other categories. Read the website.
3) This benchmark is embarrassingly threaded (probably runs the same instance multiple times with zero interaction), so all the best scoring devices are quad core Cortex-A9s instead of dual core Kraits or even Cortex-A15s. But in most real world stuff the story is very different. Note that while Z2580 is dual core Atoms still get exaggerated bonuses on
4) 1.8GHz Exynos 4 (Galaxy Note II) scores 27598 in Antutu: http://antutulabs.com/Ranking Considering Z2580 will have SGX544MP2 probably at a pretty high clock speed it should easily have more GPU power than the 400MHz Mali-400MP4 in Exynos 4.

All of these things considered the benchmark doesn't really tell us that much, assuming it's real. At the very least we'd need to see the full breakdown of the score. Concluding from this that one one Atom = two Cortex-A15 cores is just.. I don't even really know what to call that.

If its real it has to be a Silvermont 22nm cpu . It makes perfect since that it is . As its been a year . Intel did say they were going to double moores law from 32nm threw 14nm.
So according to Intels roadmaps and what it has stated publicly this should be silvermont . IF so the power usage I am hereing in these replies is comparred to present 32nm intel atom . Which indeed should be a power hog for a PHONE! But if it is what the OP stated than its power usage can and likely is below that of medfield. If those numbers are any inducation a 4 core at 10 watts is going to be a beast . Not haswell but impressive none the less

 

[Exophase]

If its real it has to be a Silvermont 22nm cpu . It makes perfect since that it is . As its been a year . Intel did say they were going to double moores law from 32nm threw 14nm.
So according to Intels roadmaps and what it has stated publicly this should be silvermont . IF so the power usage I am hereing in these replies is comparred to present 32nm intel atom . Which indeed should be a power hog for a PHONE! But if it is what the OP stated than its power usage can and likely is below that of medfield. If those numbers are any inducation a 4 core at 10 watts is going to be a beast . Not haswell but impressive none the less

So what, you think that the Z2580 that was on Intel roadmaps and announced by Anand is just not going to be released? This is where the roadmap takes you. It also aligns with what Anand said, which is that it would be released early 2013.

Why release a new SoC if you have a superior one coming out at the same time or very shortly afterwards? No one does this.

 

[MisterMac]

This is still impressive - if it's a tweaked Medfield SoC \ Core?

Without knowing much else - it's hard but still?

Bodes well for once Silvermont actually hits.

 

[Nemesis 1]

So what, you think that the Z2580 that was on Intel roadmaps and announced by Anand is just not going to be released? This is where the roadmap takes you. It also aligns with what Anand said, which is that it would be released early 2013.

Why release a new SoC if you have a superior one coming out at the same time or very shortly afterwards? No one does this.

No one does this . Fact remains intel said thats what they are doing . I would go to the 22nm silvermont as soon as possiable . If I delayed it would be airmont that I delay not silvermont. We all know intel said they were going to rapid fire atom cpus from 32nm threw 14nm. I just going what intel has been saying along its roadmaps and so on . Intel doesn't need a 2core 32 saltwell . it needs a 22nm 4 core silvermont . foe server for phones and to stomp out Temesh. I read the at article the 4g thing on saltwell makes no sense at all . On silvermont it was in the roadmaps. Intel needs the 4g for us markets and Intel needs to sell phones to Americans

 

[Nemesis 1]

Heres what I get from the net . merrifield to be shown in feb . Intel TV will not be shown. and haswell will be shown at a latter show in the spring . So CES mmay be a bummer for intel showcase . Bummer.

 

[Exophase]

Intel never said they're releasing Merrifield at the same time as Z2580, and they never gave an early 2013 date for it. The stuff you're getting from sites is unconfirmed and probably based on faulty assumptions.

Intel does need a two core SoC in phones ASAP, not to mention something with a much better GPU, which is the whole point of Z2580. Of course a Silvermont SoC would be better but it's not ready yet.

 

[Nemesis 1]

Well I know what intel said ,, As do most people in this forum . Intel said it was doubling moores law for intel atom cpus . Its a well known fact. I do recall a push back but that was on 14nm airmont not silvermont . The 4g networking isn't a saltwell package its for Silvermont . really 2 cores on saltwell in a phone I don't believe it for 1 second . Why . Because I already think intel is good . This to me is a big step up . But intel is also eying tablets its lust in their corperate hearts. and a 4 core silvermont would be well timed and needed more so the hd4000 graphics than anything else

 

[Abwx]

Intel does need a two core SoC in phones ASAP, .

They could release a good enough one in few time but it makes
no doubt that they are sensitive to public perception and in that
respect they need a 2C that do at least 1Ghz for obvious marketing
purposes...

 

[Blandge]

Amd needs to bring a 7 watt chip to the game that BEATS Silvermont.

At 7W AMD will not be competing with Atom considering Intel already showed a public demo of Ivy Bridge at < 8W

 

[Nemesis 1]

I think that was Haswell running at below 8 watts . I believe utube can help you out on that demo video . IB at 10 watts is impressive . So ya haswell will likely be in the 7 watt range . Only because of SOC and SOIX. But if you look at haswell cpu the power useage claimed by cpu is up over IB. But for the whole system it should be down as components left the mb and are on die now.

Haswell isn't intended as a tablet cpu , The fact that it can reach into that market gives Intel the highend , Thats nice but also expensive. Intel needs a good tablet cpu that will last 2 years until airmont is released

 

[MightyMalus]

Out of all the benchmarks... these ones? Pssh, I call BS.

And then they compare it with Cortex A9 chips...lawl. Exynos 5 been out for awhile and SnappyDragon with Adreno 320 also. And then, these benchmarks? Out of all the other ones...oh, wait. No wonder! Its GPU, its...its about to be 3 years old. And single core? ugh...wtf are they thinking?

What ever this Intel chip actually is, it won't be any good even for "smartphones".

By the time Intel gets down to 14nm, a lot of things will happen, the most "interesting" probably coming from Nvidia. 2013 is all Tegra 4, Exynos 5 Quad, Snappy S4 Prime and whatever Apple makes. 2014 will be interesting.

 

[Homeles]

Out of all the benchmarks... these ones? Pssh, I call BS.

And then they compare it with Cortex A9 chips...lawl. Exynos 5 been out for awhile and SnappyDragon with Adreno 320 also. And then, these benchmarks? Out of all the other ones...oh, wait. No wonder! Its GPU, its...its about to be 3 years old. And single core? ugh...wtf are they thinking?

What ever this Intel chip actually is, it won't be any good even for "smartphones".

By the time Intel gets down to 14nm, a lot of things will happen, the most "interesting" probably coming from Nvidia. 2013 is all Tegra 4, Exynos 5 Quad, Snappy S4 Prime and whatever Apple makes. 2014 will be interesting.

They're random benchmarks from an unreleased chip. Do you really think you've stumbled upon something profound?

 

[Nemesis 1]

Its likely the Z2580 . Google that and its really screwy. Some are saying its 14nm lol '

I did find 2 things of interest . There is a Z2580 L that goes in phones it 2 cpu cores and 2 gpu cores. The Z2580 W is for windows tablets and up . The funny part its 2cpu cores and 1 gpu core . Doesn't make sense unles its IB type gpu which does make sense.

Also in AT article on test the power draw of these atom and arm processors .
He had a hard time measuring the full power draw of qaulcomms cpu because of this

the CPU specific numbers for Qualcomm exclude any power consumed by the L2 cache. The total platform power numbers do include it however as they are measured at the battery.

The Z2580 has done the same thing as qualcomm

 

[Exophase]

It's pointless to try to evaluate a GPU based on the number of cores it has, or even worse based on whatever the manufacturer feels like calling cores. It's totally different from multi-core on a CPU.