Samsung Exynos Thread (big.LITTLE Octa-core)

[geoxile]

Qualcomm looks completely out of the Galaxy S6 from an AP standpoint.

Expect Exynos + MDM9x35 in USA and Exynos + whatever-else in other regions.

Doesn't Samsung have their own LTE radio IP? Not as cutting edge as Intel's discrete offerings or what Qualcomm packs but probably good enough that no one will notice

 

[Arachnotronic]

Doesn't Samsung have their own LTE radio IP? Not as cutting edge as Intel's discrete offerings or what Qualcomm packs but probably good enough that no one will notice

Need Qualcomm for CDMA (Verizon and Sprint). Could probably get away with their own LTE radio or an Intel one in the AT&T/T-Mobile versions.

 

[geoxile]

Don't know about Sprint but didn't Verizon want to drop CDMA for future phones?

 

[Arachnotronic]

Don't know about Sprint but didn't Verizon want to drop CDMA for future phones?

It wants to, but not sure if LTE-only is feasible today.

 

[el etro]

Qualcomm sleeped so hard with their main core IP. Is very good that another player in the ARM game can match Apple on per-thread(because single-thread don't mean much about the total performance of the chip) performance.

 

[Roland00Address]

Don't know about Sprint but didn't Verizon want to drop CDMA for future phones?

The day that they set for this is now in 2016. The plan was originally 2015 for LTE only phones (using Voice over LTE) but that was moved back in mid 2014. Now we do have LTE only modems, hotspots, and tablets but those are non voice devices.

So in effect Qualcomm is still going to be dominant in the modem space for all of 2015 in the US market. Who knows in international markets due to GSM. Who knows in 2016.

 

[imported_ats]

Why do you say this?

(I'm not arguing, I just would appreciate a detailed view as to why you believe so)

Because the way to generate the multi-core result doesn't have any relation to any actual multi-context applications in the phone/tablet markets. At best you could make the argument that the multi-core result could be applicable to the bulk server market, but in that case you would actually start with a decent benchmark like SpecCPU 2006.

 

[imported_ats]

Qualcomm sleeped so hard with their main core IP. Is very good that another player in the ARM game can match Apple on per-thread(because single-thread don't mean much about the total performance of the chip) performance.

Um, in phone and tablet, single threat is basically the only thing that matters! Multi-thread really isn't a useful attribute, esp after you hit 2 hardware contexts. 4 and 8 core phones/tablets are 100% marketing BS.

 

[III-V]

Qualcomm looks completely out of the Galaxy S6 from an AP standpoint.

Expect Exynos + MDM9x35 in USA and Exynos + whatever-else in other regions.

They could even go with Intel's LTE modem... not sure what the power consumption hit would be, though.

 

[Nothingness]

Because the way to generate the multi-core result doesn't have any relation to any actual multi-context applications in the phone/tablet markets. At best you could make the argument that the multi-core result could be applicable to the bulk server market, but in that case you would actually start with a decent benchmark like SpecCPU 2006.

SPEC 2006 isn't a decent multi core benchmark (at least no much better than Geekbench in the way it uses multiple cores), and isn't a decent server benchmark either

 

[imported_ats]

SPEC 2006 isn't a decent multi core benchmark (at least no much better than Geekbench in the way it uses multiple cores), and isn't a decent server benchmark either

For the types of batch workloads run on batch/bulk servers its a pretty good analog, and its actual components are significantly better than the poor kernels used in geekbench. Its certainly significantly more appropriate to the actual workloads than anything geekbench.

 

[SAAA]

Um, in phone and tablet, single threat is basically the only thing that matters! Multi-thread really isn't a useful attribute, esp after you hit 2 hardware contexts. 4 and 8 core phones/tablets are 100% marketing BS.

Let me tell you first that I agree with this comment 99%, but there can be room for doubts so... in which conditions could those 4 core in mobile be useful in any way: games, other heavy apps?

And if the answer is games only, how many of those load more than 1 core, or better how many get a significant improvement like on PCs?
I always thought GPU was far more important in such games, especially mobile ones were the logic and AI isn't anything special at all.

 

[Nothingness]

For the types of batch workloads run on batch/bulk servers its a pretty good analog, and its actual components are significantly better than the poor kernels used in geekbench. Its certainly significantly more appropriate to the actual workloads than anything geekbench.

I am still looking for multi-threaded tasks in SPEC CPU, or for DB tasks, or for Web servers tasks, or for JIT-ed code. Can you point me to these server tasks in SPEC CPU, or to the parts of SPEC CPU that are somewhat related?

This is what SPEC says:

SPEC CPU2006 focuses on compute intensive performance

This at best makes it a HPC benchmark (and even then it lacks multi-threading with data sharing benchmarks which matter a lot).

I'm not discussing Geekbench vs SPEC CPU, both have their uses. But SPEC CPU isn't a server benchmark.

 

[Sweepr]

They could even go with Intel's LTE modem... not sure what the power consumption hit would be, though.

The Galaxy Note 4 already packs an Intel XMM7260 modem (N910U Exynos version). I don't think there is a significant difference in battery life, if anything it might last a little longer than the N910C Exynos version (based on the Ericsson Thor M7450 modem) though that's speculation on my part.

Doesn't Samsung have their own LTE radio IP? Not as cutting edge as Intel's discrete offerings or what Qualcomm packs but probably good enough that no one will notice

It does, Korean Galaxy Note 4 for example uses a 20nm Shannon 303 CAT6 modem. Their recently announced Galaxy Note 4 S-LTE (previously rumoured to be the first Snapdragon S810 phone) is powered by Exynos 7 Octa + 20nm Shannon 333 CAT9 modem.

Edit: US Galaxy S6 versions will most likely pack Exynos + Qualcomm modem (due to Verizon/Sprint CDMA), they have done this in the past with the Galaxy Note 2 and the International Galaxy S3.

 

[imported_ats]

I am still looking for multi-threaded tasks in SPEC CPU, or for DB tasks, or for Web servers tasks, or for JIT-ed code. Can you point me to these server tasks in SPEC CPU, or to the parts of SPEC CPU that are somewhat related?

This is what SPEC says:
This at best makes it a HPC benchmark (and even then it lacks multi-threading with data sharing benchmarks which matter a lot).

I'm not discussing Geekbench vs SPEC CPU, both have their uses. But SPEC CPU isn't a server benchmark.

A fairly significant amount of servers are sold to do bulk/batch processing and SpecRate measures that fairly accurately. In addition, certain subtests of SpecRate tend to be useful proxies for performance in DB intensive workloads(specifically SpecRate_GCC).

And SpecRate is certainly used as a server CPU benchmark both by buyers and by designers of CPUs. The primary benchmarks that are used when designing server cpus are SpecRate, TPC, and an effective analog of the various web/java server benchmarks.

 

[Nothingness]

We are getting very off topic here. As far as gcc goes, I agree it's the best out of all SPEC CPU, but for instance its I cache pressure is miles away from what DB achieve (or fail to achieve ).

The CPU team I work in certainly uses SPEC but we're not designing server chips. Other benchmarks are used by the team evaluating server tasks.

 

[Sweepr]

Tech Report just posted a very interesting article about Exynos 5433 (Exynos 7 Octa):
From my own testing I found their CPU/GPU benchmark scores quite low, but their technical investigation about Cortex A57/A53 and Mali T760 is definitely worth a read.

Finally some info about Exynos 5433's die size:

Exynos 5433 (20nm): 113mm²
Apple A8 (20nm): 89mm²

4x Cortex A57 cores: 15.1mm²
2x Enhanced Cyclone cores: 24.4mm² (2x 12.2mm²)

Four Cortex A57 cores are only slightly bigger than a single Enhanced Cyclone core from Apple A8 (both 20nm). Eight Cortex A53 cores are smaller than a single Enhanced Cyclone core.

The A53's microarchitecture borrows heavily from the Cortex-A7 before it. The A53 can issue two instructions per clock cycle, and instructions execute in program order. The main execution pipeline is just five stages long, while the floating-point/SIMD side has seven stages. ARM thinks the A53 has taken this simple structure about as far as possible in terms of instruction throughput and power efficiency. Thanks to a host of tweaks—including better branch prediction, higher internal throughput, and power reductions than can be converted back into performance—the A53 is over 40% faster than the Cortex A7, according to ARM's own estimates. (In fact, ARM tells us the A53 is roughly 15% faster than the mid-sized Cortex-A9 rev4.) Crucially, the Cortex-A53 is fully ARMv8 and 64-bit compliant.

The Cortex-A57, meanwhile, is ARM's largest core. Derived from the Cortex-A15 used in a number of today's phones and tablets, the A57 adds ARMv8 support and incorporates a number of changes meant to increase instruction throughput. ARM intends to see the A57 used in servers, not just mobile devices, so it's pretty beefy. This core can fetch, decode, and dispatch three instructions per clock cycle. The engine gets wider after that, as illustrated in the block diagram above, and it executes instructions out of program order to improve throughput. The A57 is quad-issue into the integer execution units, dual-issue to the floating-point/SIMD units, and dual-issue to the load/store unit. ARM estimates the A57 outperforms the A15 by 25% or better. A single A57 cluster can host up to four CPU cores, and those cores use a single, shared L2 cache up to 2MB in size.

Earlier SoCs have deployed big.LITTLE in relatively simple fashion, swapping threads between a pair of quad-core clusters or migrating directly between big and little cores as needed. More recently, ARM and its partners have moved toward an arrangement known as global task scheduling in order to extract the most efficiency out of big.LITTLE operation. Global task scheduling is a form of asymmetrical multiprocessing in which all cores are active. The OS scheduler—in this case, a modified version of the Android kernel—chooses where to place threads. Newer Exynos SoCs, including the 5433, have been widely reported to use global task scheduling.

...The Exynos 5433 has another major piece of ARM tech inside: the CoreLink CCI-400 north bridge, which glues together the CPU clusters, the Mali graphics processor, and everything else. ARM's north bridges support the proper interfaces and provide hardware cache coherency, so they should work seamlessly with big.LITTLE thread migration.

Midgard is a bit of an unconventional GPU architecture in terms of its execution model, but it has a robust feature set, especially for the mobile space. The Mali-T760 supports 64-bit addressing and can handle IEEE-compliant floating-point datatypes, including 64-bit double precision. In fact, the Mali-T760 has a nearly desktop-class feature set, with support for DirectX 11.1 (feature level 11_1, or the real thing), OpenGL ES 3.1, and OpenCL 1.2.

Thanks to this combination of high mathematical precision and standards compliance, the T760 is perhaps better suited for GPU computing than most of its competition in the mobile GPU market. ARM was one of the first companies to join the HSA consortium, along with AMD, and has publicly supported that effort.

...The fact that the Note 4 reaches nearly six hours of run time in our gaming test, with Epic Citadel, is an indicator that its Mali-T760 GPU is reasonably power-efficient, too.

Speaking of efficiency, our battery life results suggest that both the Cortex A57/A53 big.LITTLE CPU config and the Mali-T760 MP6 graphics processor in the Exynos 5433 are reasonably power efficient when paired with Samsung's 20-nm fab process. At the very least, Samsung has taken the appropriate measures to ensure nice, long battery run times from the Note 4 in the scenarios we tested.

More here: http://techreport.com/review/27539/samsung-galaxy-note-4-with-the-exynos-5433-processor

I will also post some of my updated benchmark results and comparisons (Cortex A57 vs A15, Exynos vs Snapdragon/A8, etc.) tomorrow.

 

[Sweepr]

New/Updated Exynos 7 Octa (Exynos 5433) Benchmark Results - Part 1

Geekbench 3

Exynos 5433 @ 32-bit

Detailed result: http://browser.primatelabs.com/geekbench3/1845456

Exynos 5420 (Samsung Galaxy Note 3) @ 32-bit - 4x Cortex A15 @ 1.9GHz/ 4x Cortex A7 @ 1.3GHz
Multi-Core: 2960
Single-Core: 950
Detailed result: http://browser.primatelabs.com/geekbench3/1439222

Apple A8 @ 64-bit (Apple iPhone 6) - 2x Enhanced Cyclone cores @ 1.4GHz
Multi-Core: 2925
Single-Core: 1630
Detailed result: http://browser.primatelabs.com/geekbench3/1440692

Snapdragon 805 @ 32-bit (Samsung Galaxy Note 4 - N190G) - 4x Krait cores @ 2.7GHz
Multi-Core: 3220
Single-Core: 1095
Detailed results: http://browser.primatelabs.com/geekbench3/1723196

Snapdragon 810 (LG G Flex 2) @ 64-bit - 4x Cortex A57 @ 2GHz / 4x Cortex @ 1.5GHz
Multi-Core: 4564
Single-Core: 1255
Detailed result: http://browser.primatelabs.com/geekbench3/1714750

Tegra K1 Denver (Google/HTC Nexus 9) @ 64-bit - 2x Denver cores @ 2.3GHz
Multi-Core: 3580
Single-Core: 2080
Detailed result: http://browser.primatelabs.com/geekbench3/1199558

Tegra K1 (NVIDIA Shield Tablet) @ 32-bit - 4x Cortex A15
Multi-Core: 3590
Single-Core: 1105
Detailed result: http://browser.primatelabs.com/geekbench3/1807558

Exynos 7420 @ 64-bit (Samsung Galaxy S6) - 4x Cortex A57 @ 2.1GHz / 4x Cortex A53 @ 1.5GHz (unconfirmed clockspeeds)
Multi-Core: 5478
Single-Core: 1520
Detailed result: http://browser.primatelabs.com/geekbench3/1780313

VoluMark

The VoluMark benchmark is based on the Volumize volume rendering engine.

VoluMark is based on pure CPU rendering without using the GPU. It is highly optimized with SIMD and multi-threading and an excellent tool to benchmark your CPU performance.

The benchmark has been optimized both for ARM and x86 processors.

Exynos 5433

Measured VoluMark scores are:
97 for Samsung ARM A57 Exynos 5433 Quad 1.9 Ghz
80 for Samsung ARM A15 Exynos 5420 Quad 1.9 Ghz
72 for Nvidia Denver K1 Dual 2.5 Ghz
58 for Qualcomn 805 Krait Quad 2.7 Ghz
58 for Intel Bay Trail Z3745 Quad 1.86 Ghz
55 for Qualcomn 801 Krait Quad 2.5 Ghz
45 for Qualcomn 600 Krait Quad 1.9 Ghz
43 for QualComn 410 ARM A53 Quad 1.2 Ghz
29 for Nvidia Tegra 3 ARM A9 Quad 1.2 Ghz
26 for QualComn 400 ARM A7 Quad 1.2 Ghz
22 for RockChip RK3066 ARM A9 Dual 1.6 Ghz
19 for QualComn 400 Krait Dual 1.4 Ghz

If we look at Exynos 5433 and Exynos 5420 the A57/A53 combination is ~21% faster than A15/A7 running at the same clockspeed. Thanks to increased IPC the 1.9/1.3GHz A57/A53 combination also beats Mediatek MT6595's A17/A7 cores @ 2.2/1.7GHz (score: 93).

https://play.google.com/store/apps/details?id=volumize.be

Quadrant

Exynos 5433

Quad A57 + Quad A53 is 28.5% faster than Quad A15 + Quad A7 @ Quadrant (1.9/1.3GHz Exynos 5433 vs 1.9/1.3GHz Exynos 5420).

CPU Prime

Exynos 5433

Exynos 5422 (Samsung Galaxy S5 Exynos)

A57/A53 provides 22.7% better performance/clock than A15/A7 in this benchmark (Exynos 5433 vs 5422 @ 1.9/1.3GHz).

Snapdragon 801 (Sony Zperia Z3 Compact Tablet)

CF-Bench

Exynos 5433

Exynos 5422 (Samsung Galaxy S5 Exynos)

A57/A53 is 18.5% faster than A15/A7 @ CF-Bench (Exynos 5433 vs 5422 @ 1.9/1.3GHz).

Snapdragon 805 (Nexus 6) and Tegra K1 Denver (Nexus 9)

What ARM has to say about Cortex A57/A53:

The Cortex-A15 and Cortex-A7 cores represent the first generation of big.LITTLE hardware. ARM has announced 2 new CPU cores that are also capable of big.LITTLE processing, the Cortex-A57 and the Cortex-A53 processors. The Cortex-A57 processor is a performance optimized CPU, bringing more than 25% more performance per clock cycle, higher frequency capability, and slightly higher efficiency than the Cortex-A15 processor. The Cortex-A53 processor is a LITTLE core with 40% more performance per clock cycle, at about the same power efficiency as Cortex-A7 processor.

http://community.arm.com/docs/DOC-2875

Based on the test results above (excluding Geekbench's +50% performance uplift) big.LITTLE Cortex A57+A53 was ~23% faster (overall) than Cortex A15+A7 operating at the same frequency.

As a side note, this is an exceptional device to run emulators. I've tested NES, SNES, N64, GC/GBA, PSX, PSP, Neo Geo, Mega Drive, Dreamcast and a few other Android emulators. Even GC/Wii games are quite playable through the demanding GC/Wii Dolphin emulator, most of them are running at >15 FPS (it requires some patience though; Exynos 7420's +20% CPU performance bump should greatly help here). Being able to emulate generation 6 console games on a phone is an impressive feat IMHO.

That's it for today, Part 2 tomorrow.

 

[lopri]

My Fire phone gets 74 for that Volumark thing and Z Ultra gets 63. Consistently on both so I thought the scores are affected by resolution. (Fire phone = 720p, Ultra Z = 1080p) Both phones are running on S800/2.20 GHz.

 

[Andrei.]

The TechReport piece has some mistakes in it. I hope to have my article up here on Monday.

 

[Idontcare]

The TechReport piece has some mistakes in it. I hope to have my article up here on Monday.

Looking forward to it!

New/Updated Exynos 7 Octa (Exynos 5433) Benchmark Results - Part 1

Geekbench 3

Exynos 5433 @ 32-bit

Detailed result: http://browser.primatelabs.com/geekbench3/1845456

Exynos 5420 (Samsung Galaxy Note 3) @ 32-bit - 4x Cortex A15 @ 1.9GHz/ 4x Cortex A7 @ 1.3GHz
Multi-Core: 2960
Single-Core: 950
Detailed result: http://browser.primatelabs.com/geekbench3/1439222

Apple A8 @ 64-bit (Apple iPhone 6) - 2x Enhanced Cyclone cores @ 1.4GHz
Multi-Core: 2925
Single-Core: 1630
Detailed result: http://browser.primatelabs.com/geekbench3/1440692

Snapdragon 805 @ 32-bit (Samsung Galaxy Note 4 - N190G) - 4x Krait cores @ 2.7GHz
Multi-Core: 3220
Single-Core: 1095
Detailed results: http://browser.primatelabs.com/geekbench3/1723196

Snapdragon 810 (LG G Flex 2) @ 64-bit - 4x Cortex A57 @ 2GHz / 4x Cortex @ 1.5GHz
Multi-Core: 4564
Single-Core: 1255
Detailed result: http://browser.primatelabs.com/geekbench3/1714750

Tegra K1 Denver (Google/HTC Nexus 9) @ 64-bit - 2x Denver cores @ 2.3GHz
Multi-Core: 3580
Single-Core: 2080
Detailed result: http://browser.primatelabs.com/geekbench3/1199558

Tegra K1 (NVIDIA Shield Tablet) @ 32-bit - 4x Cortex A15
Multi-Core: 3590
Single-Core: 1105
Detailed result: http://browser.primatelabs.com/geekbench3/1807558

Exynos 7420 @ 64-bit (Samsung Galaxy S6) - 4x Cortex A57 @ 2.1GHz / 4x Cortex A53 @ 1.5GHz (unconfirmed clockspeeds)
Multi-Core: 5478
Single-Core: 1520
Detailed result: http://browser.primatelabs.com/geekbench3/1780313

VoluMark



Exynos 5433

Measured VoluMark scores are:
97 for Samsung ARM A57 Exynos 5433 Quad 1.9 Ghz
80 for Samsung ARM A15 Exynos 5420 Quad 1.9 Ghz
72 for Nvidia Denver K1 Dual 2.5 Ghz
58 for Qualcomn 805 Krait Quad 2.7 Ghz
58 for Intel Bay Trail Z3745 Quad 1.86 Ghz
55 for Qualcomn 801 Krait Quad 2.5 Ghz
45 for Qualcomn 600 Krait Quad 1.9 Ghz
43 for QualComn 410 ARM A53 Quad 1.2 Ghz
29 for Nvidia Tegra 3 ARM A9 Quad 1.2 Ghz
26 for QualComn 400 ARM A7 Quad 1.2 Ghz
22 for RockChip RK3066 ARM A9 Dual 1.6 Ghz
19 for QualComn 400 Krait Dual 1.4 Ghz

If we look at Exynos 5433 and Exynos 5420 the A57/A53 combination is ~21% faster than A15/A7 running at the same clockspeed. Thanks to increased IPC the 1.9/1.3GHz A57/A53 combination also beats Mediatek MT6595's A17/A7 cores @ 2.2/1.7GHz (score: 93).

https://play.google.com/store/apps/details?id=volumize.be

Quadrant

Exynos 5433

Quad A57 + Quad A53 is 28.5% faster than Quad A15 + Quad A7 @ Quadrant (1.9/1.3GHz Exynos 5433 vs 1.9/1.3GHz Exynos 5420).

CPU Prime

Exynos 5433

Exynos 5422 (Samsung Galaxy S5 Exynos)

A57/A53 provides 22.7% better performance/clock than A15/A7 in this benchmark (Exynos 5433 vs 5422 @ 1.9/1.3GHz).

Snapdragon 801 (Sony Zperia Z3 Compact Tablet)

CF-Bench

Exynos 5433

Exynos 5422 (Samsung Galaxy S5 Exynos)

A57/A53 is 18.5% faster than A15/A7 @ CF-Bench (Exynos 5433 vs 5422 @ 1.9/1.3GHz).

Snapdragon 805 (Nexus 6) and Tegra K1 Denver (Nexus 9)

What ARM has to say about Cortex A57/A53:



http://community.arm.com/docs/DOC-2875

Based on the test results above (excluding Geekbench's +50% performance uplift) big.LITTLE Cortex A57+A53 was ~23% faster (overall) than Cortex A15+A7 operating at the same frequency.

As a side note, this is an exceptional device to run emulators. I've tested NES, SNES, N64, GC/GBA, PSX, PSP, Neo Geo, Mega Drive, Dreamcast and a few other Android emulators. Even GC/Wii games are quite playable through the demanding GC/Wii Dolphin emulator, most of them are running at >15 FPS (it requires some patience though; Exynos 7420's +20% CPU performance bump should greatly help here). Being able to emulate generation 6 console games on a phone is an impressive feat IMHO.

That's it for today, Part 2 tomorrow.

OK, so you do reviews in a way that really makes me want to read them and enjoy reading them.

Please continue!

 

[krumme]

New/Updated Exynos 7 Octa (Exynos 5433) Benchmark Results - Part 1

Geekbench 3

Exynos 5433 @ 32-bit

Detailed result: http://browser.primatelabs.com/geekbench3/1845456

Exynos 5420 (Samsung Galaxy Note 3) @ 32-bit - 4x Cortex A15 @ 1.9GHz/ 4x Cortex A7 @ 1.3GHz
Multi-Core: 2960
Single-Core: 950
Detailed result: http://browser.primatelabs.com/geekbench3/1439222

Apple A8 @ 64-bit (Apple iPhone 6) - 2x Enhanced Cyclone cores @ 1.4GHz
Multi-Core: 2925
Single-Core: 1630
Detailed result: http://browser.primatelabs.com/geekbench3/1440692

Snapdragon 805 @ 32-bit (Samsung Galaxy Note 4 - N190G) - 4x Krait cores @ 2.7GHz
Multi-Core: 3220
Single-Core: 1095
Detailed results: http://browser.primatelabs.com/geekbench3/1723196

Snapdragon 810 (LG G Flex 2) @ 64-bit - 4x Cortex A57 @ 2GHz / 4x Cortex @ 1.5GHz
Multi-Core: 4564
Single-Core: 1255
Detailed result: http://browser.primatelabs.com/geekbench3/1714750

Tegra K1 Denver (Google/HTC Nexus 9) @ 64-bit - 2x Denver cores @ 2.3GHz
Multi-Core: 3580
Single-Core: 2080
Detailed result: http://browser.primatelabs.com/geekbench3/1199558

Tegra K1 (NVIDIA Shield Tablet) @ 32-bit - 4x Cortex A15
Multi-Core: 3590
Single-Core: 1105
Detailed result: http://browser.primatelabs.com/geekbench3/1807558

Exynos 7420 @ 64-bit (Samsung Galaxy S6) - 4x Cortex A57 @ 2.1GHz / 4x Cortex A53 @ 1.5GHz (unconfirmed clockspeeds)
Multi-Core: 5478
Single-Core: 1520
Detailed result: http://browser.primatelabs.com/geekbench3/1780313

VoluMark



Exynos 5433

Measured VoluMark scores are:
97 for Samsung ARM A57 Exynos 5433 Quad 1.9 Ghz
80 for Samsung ARM A15 Exynos 5420 Quad 1.9 Ghz
72 for Nvidia Denver K1 Dual 2.5 Ghz
58 for Qualcomn 805 Krait Quad 2.7 Ghz
58 for Intel Bay Trail Z3745 Quad 1.86 Ghz
55 for Qualcomn 801 Krait Quad 2.5 Ghz
45 for Qualcomn 600 Krait Quad 1.9 Ghz
43 for QualComn 410 ARM A53 Quad 1.2 Ghz
29 for Nvidia Tegra 3 ARM A9 Quad 1.2 Ghz
26 for QualComn 400 ARM A7 Quad 1.2 Ghz
22 for RockChip RK3066 ARM A9 Dual 1.6 Ghz
19 for QualComn 400 Krait Dual 1.4 Ghz

If we look at Exynos 5433 and Exynos 5420 the A57/A53 combination is ~21% faster than A15/A7 running at the same clockspeed. Thanks to increased IPC the 1.9/1.3GHz A57/A53 combination also beats Mediatek MT6595's A17/A7 cores @ 2.2/1.7GHz (score: 93).

https://play.google.com/store/apps/details?id=volumize.be

Quadrant

Exynos 5433

Quad A57 + Quad A53 is 28.5% faster than Quad A15 + Quad A7 @ Quadrant (1.9/1.3GHz Exynos 5433 vs 1.9/1.3GHz Exynos 5420).

CPU Prime

Exynos 5433

Exynos 5422 (Samsung Galaxy S5 Exynos)

A57/A53 provides 22.7% better performance/clock than A15/A7 in this benchmark (Exynos 5433 vs 5422 @ 1.9/1.3GHz).

Snapdragon 801 (Sony Zperia Z3 Compact Tablet)

CF-Bench

Exynos 5433

Exynos 5422 (Samsung Galaxy S5 Exynos)

A57/A53 is 18.5% faster than A15/A7 @ CF-Bench (Exynos 5433 vs 5422 @ 1.9/1.3GHz).

Snapdragon 805 (Nexus 6) and Tegra K1 Denver (Nexus 9)

What ARM has to say about Cortex A57/A53:



http://community.arm.com/docs/DOC-2875

Based on the test results above (excluding Geekbench's +50% performance uplift) big.LITTLE Cortex A57+A53 was ~23% faster (overall) than Cortex A15+A7 operating at the same frequency.

As a side note, this is an exceptional device to run emulators. I've tested NES, SNES, N64, GC/GBA, PSX, PSP, Neo Geo, Mega Drive, Dreamcast and a few other Android emulators. Even GC/Wii games are quite playable through the demanding GC/Wii Dolphin emulator, most of them are running at >15 FPS (it requires some patience though; Exynos 7420's +20% CPU performance bump should greatly help here). Being able to emulate generation 6 console games on a phone is an impressive feat IMHO.

That's it for today, Part 2 tomorrow.

Excellent keep em coming.
The A53 volumark score seems very high to me especially vs A7. Why is that?

 

[III-V]

Excellent keep em coming.
The A53 volumark score seems very high to me especially vs A7. Why is that?

Because it's an ARM A7, not an Apple A7

 

[Lepton87]

There are still going to be people who will claim it is still inferior to Apple SOC because it has better ST score and any result using more than X threads don't matter and is pure BS.

X - the number of threads Apple AX chip currently can handle which is 2 for a phone and 3 for a tablet. When apple releases a chip with more threads this will change, because software will only be ready when apple is.

 

[lopri]

Is it necessary to quote a whole post with 10+ images.. some of us are accessing this forum on mobile devices.