samsung caught boosting exynos benchmarks

[poofyhairguy]

Maybe this is why they dumped PowerVR in the more recent Exynos for Mali.

 

[Staples]

LOL at people basing their buying decision on benchmarks alone

No one uses them more thanl Apple haters. They love to use benchmarks as a prime example of why the iPhone is inferior.

 

[Mopetar]

Maybe this is why they dumped PowerVR in the more recent Exynos for Mali.

Huh?

They haven't used PowerVR in years. They've used the Mali GPU ever since they started making dual-core chips. I think the only reason they ever used PowerVR to start with was because they were making the chips for the iPhone and Apple may have wanted them to use PowerVR.

 

[dagamer34]

Huh?

They haven't used PowerVR in years. They've used the Mali GPU ever since they started making dual-core chips. I think the only reason they ever used PowerVR to start with was because they were making the chips for the iPhone and Apple may have wanted them to use PowerVR.

Exynos 5420 uses PowerVR SGX 544MP3 @ 533Mhz in the Korean Galaxy S4. Also, pre-iPhone 4, the SoCs used in Apple phones have Samsung part numbers, they are 100% Samsung designs.

 

[poofyhairguy]

Huh?

They haven't used PowerVR in years. They've used the Mali GPU ever since they started making dual-core chips. I think the only reason they ever used PowerVR to start with was because they were making the chips for the iPhone and Apple may have wanted them to use PowerVR.

http://www.sammobile.com/2013/07/23...-5-octa-soc-based-on-arm-mali-t628-mp6-cores/

 

[Eug]

This is extremely lame on Samsung's part. Extremely. Something I'd expect from some fly-by-night little company.

P.S. I'm thinking this should be a front and centre article, on deceptive business practices, not some sidebar.

 

[stormkroe]

Hmm... looks like somebody already noticed this...
Not sure if it was happening the same way back then but I sure endured a lot of ridicule (mostly from other sites).
I'm just super skeptical after surviving the PC GPU debacles every other week.

 

[lopri]

P.S. I'm thinking this should be a front and centre article, on deceptive business practices, not some sidebar.

Samsung is reportedly warming up to Intel for future products. So there is that. :sneaky: It's not like this site published anything about recent allegation about Intel's shady tactics. At least one dude seems to have axe to grind against Samsung so it is being reported.

My stance hasn't changed with regard to the lameness and unethical nature of this latest scandal of Samsung, but it is a tad bit softened after learning the discrepancy is 53 MHz. (480 MHz v. 533 MHz) I initially thought it's 400 MHz v. 533 MHz. It is ridiculous still that Samsung would stoop so low for such minuscule PR.

 

[tvdang7]

Some people have alot of time on their hands.

 

[Germanic]

I really don't see this thread going anywhere...

 

[lopri]

Some people have alot of time on their hands.

You may want to check out CPU forum or GPU forum. You'll be surprised to see how much free time folks have.

 

[Renob]

http://www.engadget.com/2013/07/30/samsung-benchmarks/

not surprised. they're a shady company.

lol they are ALL shady!!!

 

[BlahBlahYouToo]

I really don't see this thread going anywhere...

well, you're korean so...

 

[WelshBloke]

I still don't understand the point of this. If the GPU can run at those speeds under stress running a benchmark why not let it ramp up to there any time?

 

[grkM3]

I still don't understand the point of this. If the GPU can run at those speeds under stress running a benchmark why not let it ramp up to there any time?

Maybe Samsung was trying to match real world performance to the snapdragon gs4?

There is a 1.6ghz and 1.8ghz CPU exynos soc also

 

[Suspicious-Teach8788]

Maybe Samsung was trying to match real world performance to the snapdragon gs4?

There is a 1.6ghz and 1.8ghz CPU exynos soc also

It's not even that. What about battery life? What general UI tasks require a fully clocked GPU?

Isn't this like how your AMD or NVidia card will run at reduced clock speeds in your OS but will ramp up for gaming?

I don't see how this is an issue. If anything, this is like what kernel devs do.... They make sure that in general tasks, swiping, loading webpages, your CPU won't crank up to 4x1.5ghz or whatever. It only does so in gaming or actual CPU intensive tasks.

 

[grkM3]

It's not even that. What about battery life? What general UI tasks require a fully clocked GPU?

Isn't this like how your AMD or NVidia card will run at reduced clock speeds in your OS but will ramp up for gaming?

I don't see how this is an issue. If anything, this is like what kernel devs do.... They make sure that in general tasks, swiping, loading webpages, your CPU won't crank up to 4x1.5ghz or whatever. It only does so in gaming or actual CPU intensive tasks.

Battery could also be a part of it but honestly how much is 40mhz going to save? And why are people making a huge deal of this anyways? Its hitting its speeds as advertised in what counts the most,benchmarks to brag about.

My gs4 gets super hot while playing candy crush so they might have tamed the GPU down for battery life but still why such a small change.

 

[WelshBloke]

It's not even that. What about battery life? What general UI tasks require a fully clocked GPU?

Isn't this like how your AMD or NVidia card will run at reduced clock speeds in your OS but will ramp up for gaming?

I don't see how this is an issue. If anything, this is like what kernel devs do.... They make sure that in general tasks, swiping, loading webpages, your CPU won't crank up to 4x1.5ghz or whatever. It only does so in gaming or actual CPU intensive tasks.

Well as long as it's hitting those speeds under full load I don't see an issue either.

I was understanding that it was only hitting the higher speeds in benchmarking and not in games when you need it, which is disingenuous at the least and dishonest at the worst.

 

[Suspicious-Teach8788]

Well as long as it's hitting those speeds under full load I don't see an issue either.

I was understanding that it was only hitting the higher speeds in benchmarking and not in games when you need it, which is disingenuous at the least and dishonest at the worst.

You're right its hitting higher speeds in benchmarking. 533 vs 480 for games, but to me big deal? That's like saying the phone can't continue long term for at 533, so they clocked at 480.

I dunno, it's like Ivy Bridge. The turbo clocks are only hit in the beginning, but after prolonged use the clocks settle down between the base and the turbo to maintain a TDP....

I guess I just don't see it as big of a deal, and I value the battery and heat savings that Samsung's implementing. If the GS4 and other phones had no heat and throttling issues, then I'd be more annoyed because its clear the phone can run at those speeds without issues, yet its just for show in benchmarks.

 

[BlahBlahYouToo]

You're right its hitting higher speeds in benchmarking. 533 vs 480 for games, but to me big deal? That's like saying the phone can't continue long term for at 533, so they clocked at 480.

I dunno, it's like Ivy Bridge. The turbo clocks are only hit in the beginning, but after prolonged use the clocks settle down between the base and the turbo to maintain a TDP....

I guess I just don't see it as big of a deal, and I value the battery and heat savings that Samsung's implementing. If the GS4 and other phones had no heat and throttling issues, then I'd be more annoyed because its clear the phone can run at those speeds without issues, yet its just for show in benchmarks.

no, it's like ivy bridge only ramping up clockspeeds only during benchmarking for the sole purpose of looking superior in comparisons.

 

[Suspicious-Teach8788]

no, it's like ivy bridge only ramping up clockspeeds only during benchmarking for the sole purpose of looking superior in comparisons.

So if you bench, you get a great score, but if you game, your ivy bridge won't ramp as high. That's already how it is.

 

[grkM3]

Its a mode the phone detects to keep the GPU pegged during a benchmark

 

[TuxDave]

I dunno, it's like Ivy Bridge. The turbo clocks are only hit in the beginning, but after prolonged use the clocks settle down between the base and the turbo to maintain a TDP....

I'm not a turbo mode expert but just some clarification. As far as I know, turbo mode isn't a flat "it's on in the beginning and turns off after some time". It's a function of power and temp and it'll drop when certain metrics are hit. If you run a 2 core benchmark on a 4 core machine, I'm pretty sure you'll stay in turbo mode for a long time.

I'm trying to be fair and turn my head sideways to see your perspective on it. To me, an opportunistic turbo mode is a useful thing. Something that turns on during benchmarks (and camera mode!) is not a useful thing. That is the first basic difference that I see.