"More to the point, Qualcomm's Snapdragon845, and Samsung's Exynos 9820, will match or exceed the KKiri 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
Interesting that you seem so confident.
You might have an answer in around five weeks from now:
"Rumors have it that for the first time Huawei may pack its own graphics processor inside it. The performance of the GPU of the Kirin 980 is speculated to be 1.5 times better than Adreno 630 that comes with Snapdragon 845."
We'll see how that plays out.
Benchmarks are one thing - sustainable performance is another thing which NOT many tech websites dig deep into, except AnandTech. AnandTech had praised the Kirin 950 on the sustained performance and thermal management 2 years back. But they criticized the recent Kirin 960 and 970. We will have to wait for a deep dive by AnandTech to see how Kirin 980 holds up.
Most definitely.
At this point all we have is rumour and speculation. I just wanted to counter what tmay has already presented as 'absolutes' (although he has absolutely nothing to back the claims up) with the only information that is really out (marking it clearly for what it is (rumours).
If they present their own GPU (something that has been heavily suspected for nearly two years now) it will be first generation and possibly lacking optimizations. On the other hand, the 7nm process will help with efficiencies overall and the second generation NPU is rumoured to be much more capable per watt than the first gen product. Added to that, one of the tentpole features of Turbo GPU is its up to 30% power saving but all that is to be seen.
There you go again, adding to the speculation in favor of your favored Huawei.
How about you just sit back and wait for the 980 to arrive in a product, and the "Turbo GT" software to be released and available for review and benchmarking?
Every major SOC maker will be a 7nm this fall, so it will be easy to compare performance amongst the new devices.
No. Speculation and rumour is all their is. Nothing more. That's why I made a point of making that clear when you jumped in feet first with your 'absolute' comments that aren't based on anything that is really known, real or independently scrutinised
You say 'this' or 'that' is this or that way but you simply don't know. No one does!
Yes, sit back and wait. That's the correct way. Mention rumours and speculation if you wish but don't start throwing absolutes around because there are none (in the context of what matters here).
Absolutely, the Kirin 970 does not do as well in benchmarks as the A11 or the Qualcomm 845; it isn't even close.
Qualcomm has it's 855 that will a a 7nm SOC coming out later than the others, but otherwise, I'm not seeing the Kirin 980 as making such a large jump with off the shelf compute cores. You may be right that it will do better than the 845 in graphics performance, but that isn't the end all be all of performance benchmarks. Hence why I stated that the competition will match or exceed the Kirin 980.
That is better from you but wasn't it you yourself who said I needed a better response?
This is what you actually said:
"More to the point, Qualcomm's Snapdragon 845, and Samsung's Exynos 9820, will match or exceed the Kirin 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
That's pretty absolute. No. "Will match or exceed..." is totally absolute!
And when I quoted you the first time (this is the second) I bolded the problem parts for a reason. Here it is:
"More to the point, Qualcomm's Snapdragon845, and Samsung's Exynos 9820, will match or exceed the Kirin 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
And then you start your explanation by bolding an entire opening paragraph on the Kirin 970!
And in your reply you throw in the Snapdragon 855 (which wasn't even mentioned) and all the absolutes are now nowhere to be seen!
What you are trying to do is re-work what you said. A simple 'my bad' would have been enough. Not even that if you hadn't thrown in the 'better response' and 'there'you go again' parts
At the end of the day you are now admitting you simply don't know. No problem with that. I don't either.
I don't agree with your lead out but as you've repackaged what you are saying to make it clear it's only your opinion, I find it more than valid.
That said, I should point out that the rumours claim the Kirin 980 will include Cortex A77 cores, which, as far as I know, do not exist. It could be a typo but what if it isn't?
How could you form an opinion on cores that no one has any information on?
Don't get me wrong, I'm inclined to think it's a typo but I won't close the door completely on that, just in case.
Okay, so here's one of the reason's that I'm confident about by statement.
First off, there is no A77 core, and the A76 core is 4 wide microarchitecture compared to Apple's current 6 wide architecture.
So this is the prediction that Anadtech has for the comparative performance of the A76 core vs the A11 Monsoon core;
"In terms of general IPC Arm promises a ~25% increase in integer workloads and a ~35% increase in ASIMD/floating point workloads. Together with up to 90% higher memory bandwidth figures compared to the A75 the A76 is then meant to provide around a 28% increase in GeekBench4 and 35% more JavaScript performance (Octane, JetStream). In AI inferencing workloads the doubled ASIMD 128-bit capabilities of the A76 serves to quadruple the general matrix multiply performance in half precision formats.
These performance figures are respectable but not quite earth-shattering considering the tone of the improvements of the µarch. However it’s to note that we’re expecting the A76 to come first be deployed in flagship SoCs on TSMC’s 7nm process which allows for increased clocks."
That pretty much describes the Kirin 980, and the Qualcomm 855, but note, the Kirin 780 is still quite far behind the Qualcomm 845 as is. The A12's microarchitecture and Apple's experience in optimization pretty much guarantees that the A12 will continue holding the crown for both single thread and multiple thread performance.
"I'm confident about by statement"
You may be as 'confident' as you see fit but your 'absolute' comments are now nowhere to be seen except for this:
That pretty much describes the Kirin 980, and the Qualcomm 855, but note, the Kirin 780 is still quite far behind the Qualcomm 845 as is.
"As is"? No. That's impossible. The Kirin 980 doesn't exist!
You are speculating, and wildly at that. "As is" has no place here. Speculation is fine though.
But if you are going to speculate why not at least find a place for those mystery Antutu benchmarks which set a new high (blasting past the 845) and the possibility of a custom GPU which is rumoured to be far faster than the GPU on the 845?
Could it be that that would blow a hole the size of the Grand Canyon in your argument?
I suggest we limit ourselves to speculation or rumours and forget about putting absolutes on the table.
After Huawei taking a huge chunk out of Portugal, Counterpoint has supposedly put Huawei ahead in Russia. Interestingly, the article claims that 70% of Apple's lost customers went to Huawei:
"More to the point, Qualcomm's Snapdragon845, and Samsung's Exynos 9820, will match or exceed the KKiri 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
Interesting that you seem so confident.
You might have an answer in around five weeks from now:
"Rumors have it that for the first time Huawei may pack its own graphics processor inside it. The performance of the GPU of the Kirin 980 is speculated to be 1.5 times better than Adreno 630 that comes with Snapdragon 845."
We'll see how that plays out.
Benchmarks are one thing - sustainable performance is another thing which NOT many tech websites dig deep into, except AnandTech. AnandTech had praised the Kirin 950 on the sustained performance and thermal management 2 years back. But they criticized the recent Kirin 960 and 970. We will have to wait for a deep dive by AnandTech to see how Kirin 980 holds up.
Most definitely.
At this point all we have is rumour and speculation. I just wanted to counter what tmay has already presented as 'absolutes' (although he has absolutely nothing to back the claims up) with the only information that is really out (marking it clearly for what it is (rumours).
If they present their own GPU (something that has been heavily suspected for nearly two years now) it will be first generation and possibly lacking optimizations. On the other hand, the 7nm process will help with efficiencies overall and the second generation NPU is rumoured to be much more capable per watt than the first gen product. Added to that, one of the tentpole features of Turbo GPU is its up to 30% power saving but all that is to be seen.
There you go again, adding to the speculation in favor of your favored Huawei.
How about you just sit back and wait for the 980 to arrive in a product, and the "Turbo GT" software to be released and available for review and benchmarking?
Every major SOC maker will be a 7nm this fall, so it will be easy to compare performance amongst the new devices.
No. Speculation and rumour is all their is. Nothing more. That's why I made a point of making that clear when you jumped in feet first with your 'absolute' comments that aren't based on anything that is really known, real or independently scrutinised
You say 'this' or 'that' is this or that way but you simply don't know. No one does!
Yes, sit back and wait. That's the correct way. Mention rumours and speculation if you wish but don't start throwing absolutes around because there are none (in the context of what matters here).
Absolutely, the Kirin 970 does not do as well in benchmarks as the A11 or the Qualcomm 845; it isn't even close.
Qualcomm has it's 855 that will a a 7nm SOC coming out later than the others, but otherwise, I'm not seeing the Kirin 980 as making such a large jump with off the shelf compute cores. You may be right that it will do better than the 845 in graphics performance, but that isn't the end all be all of performance benchmarks. Hence why I stated that the competition will match or exceed the Kirin 980.
That is better from you but wasn't it you yourself who said I needed a better response?
This is what you actually said:
"More to the point, Qualcomm's Snapdragon 845, and Samsung's Exynos 9820, will match or exceed the Kirin 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
That's pretty absolute. No. "Will match or exceed..." is totally absolute!
And when I quoted you the first time (this is the second) I bolded the problem parts for a reason. Here it is:
"More to the point, Qualcomm's Snapdragon845, and Samsung's Exynos 9820, will match or exceed the Kirin 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
And then you start your explanation by bolding an entire opening paragraph on the Kirin 970!
And in your reply you throw in the Snapdragon 855 (which wasn't even mentioned) and all the absolutes are now nowhere to be seen!
What you are trying to do is re-work what you said. A simple 'my bad' would have been enough. Not even that if you hadn't thrown in the 'better response' and 'there'you go again' parts
At the end of the day you are now admitting you simply don't know. No problem with that. I don't either.
I don't agree with your lead out but as you've repackaged what you are saying to make it clear it's only your opinion, I find it more than valid.
That said, I should point out that the rumours claim the Kirin 980 will include Cortex A77 cores, which, as far as I know, do not exist. It could be a typo but what if it isn't?
How could you form an opinion on cores that no one has any information on?
Don't get me wrong, I'm inclined to think it's a typo but I won't close the door completely on that, just in case.
Okay, so here's one of the reason's that I'm confident about by statement.
First off, there is no A77 core, and the A76 core is 4 wide microarchitecture compared to Apple's current 6 wide architecture.
So this is the prediction that Anadtech has for the comparative performance of the A76 core vs the A11 Monsoon core;
"In terms of general IPC Arm promises a ~25% increase in integer workloads and a ~35% increase in ASIMD/floating point workloads. Together with up to 90% higher memory bandwidth figures compared to the A75 the A76 is then meant to provide around a 28% increase in GeekBench4 and 35% more JavaScript performance (Octane, JetStream). In AI inferencing workloads the doubled ASIMD 128-bit capabilities of the A76 serves to quadruple the general matrix multiply performance in half precision formats.
These performance figures are respectable but not quite earth-shattering considering the tone of the improvements of the µarch. However it’s to note that we’re expecting the A76 to come first be deployed in flagship SoCs on TSMC’s 7nm process which allows for increased clocks."
That pretty much describes the Kirin 980, and the Qualcomm 855, but note, the Kirin 780 is still quite far behind the Qualcomm 845 as is. The A12's microarchitecture and Apple's experience in optimization pretty much guarantees that the A12 will continue holding the crown for both single thread and multiple thread performance.
"I'm confident about by statement"
You may be as 'confident' as you see fit but your 'absolute' comments are now nowhere to be seen except for this:
That pretty much describes the Kirin 980, and the Qualcomm 855, but note, the Kirin 780 is still quite far behind the Qualcomm 845 as is.
"As is"? No. That's impossible. The Kirin 980 doesn't exist!
You are speculating, and wildly at that. "As is" has no place here. Speculation is fine though.
But if you are going to speculate why not at least find a place for those mystery Antutu benchmarks which set a new high (blasting past the 845) and the possibility of a custom GPU which is rumoured to be far faster than the GPU on the 845?
Could it be that that would blow a hole the size of the Grand Canyon in your argument?
I suggest we limit ourselves to speculation or rumours and forget about putting absolutes on the table.
Typo; the 970 is far behind the 845.
If you are arguing that the Kirin 980 doesn't exist, then why were you fine with posting rumors to the benchmarks of that same SOC?
"Of course, AnTuTu doesn't consider CPU performance alone, so this doesn't count for much at the end of the day."
If the latest ARM core is the A76, (there is no A77 yet), and assuming that core was even available early enough for the 980 will be comparable to the 855 architecturally, both at 7nm. As for the GPU, I never stated anything about GPU performance. You might be quite correct that it exceeds both the 845 and the 855 when it arrives, and almost certainly exceeds that A12, but there is still that necessity to stay within the TDP of a smartphone, hence why I don't buy those Antutu benchmarks for the 980 which arrived before the 980 is even in production at TMSC. If you read the Anandtech article, the performance of these A76 cores is based on the theoretical high clock rate of around 3 Gnz, which isn't likely to happen for a production part.
So, I've been talking about the CPU architecture, and you are leaning on Huawei to provide a massive GPU advantage.
"More to the point, Qualcomm's Snapdragon845, and Samsung's Exynos 9820, will match or exceed the KKiri 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
Interesting that you seem so confident.
You might have an answer in around five weeks from now:
"Rumors have it that for the first time Huawei may pack its own graphics processor inside it. The performance of the GPU of the Kirin 980 is speculated to be 1.5 times better than Adreno 630 that comes with Snapdragon 845."
We'll see how that plays out.
Benchmarks are one thing - sustainable performance is another thing which NOT many tech websites dig deep into, except AnandTech. AnandTech had praised the Kirin 950 on the sustained performance and thermal management 2 years back. But they criticized the recent Kirin 960 and 970. We will have to wait for a deep dive by AnandTech to see how Kirin 980 holds up.
Most definitely.
At this point all we have is rumour and speculation. I just wanted to counter what tmay has already presented as 'absolutes' (although he has absolutely nothing to back the claims up) with the only information that is really out (marking it clearly for what it is (rumours).
If they present their own GPU (something that has been heavily suspected for nearly two years now) it will be first generation and possibly lacking optimizations. On the other hand, the 7nm process will help with efficiencies overall and the second generation NPU is rumoured to be much more capable per watt than the first gen product. Added to that, one of the tentpole features of Turbo GPU is its up to 30% power saving but all that is to be seen.
There you go again, adding to the speculation in favor of your favored Huawei.
How about you just sit back and wait for the 980 to arrive in a product, and the "Turbo GT" software to be released and available for review and benchmarking?
Every major SOC maker will be a 7nm this fall, so it will be easy to compare performance amongst the new devices.
No. Speculation and rumour is all their is. Nothing more. That's why I made a point of making that clear when you jumped in feet first with your 'absolute' comments that aren't based on anything that is really known, real or independently scrutinised
You say 'this' or 'that' is this or that way but you simply don't know. No one does!
Yes, sit back and wait. That's the correct way. Mention rumours and speculation if you wish but don't start throwing absolutes around because there are none (in the context of what matters here).
Absolutely, the Kirin 970 does not do as well in benchmarks as the A11 or the Qualcomm 845; it isn't even close.
Qualcomm has it's 855 that will a a 7nm SOC coming out later than the others, but otherwise, I'm not seeing the Kirin 980 as making such a large jump with off the shelf compute cores. You may be right that it will do better than the 845 in graphics performance, but that isn't the end all be all of performance benchmarks. Hence why I stated that the competition will match or exceed the Kirin 980.
That is better from you but wasn't it you yourself who said I needed a better response?
This is what you actually said:
"More to the point, Qualcomm's Snapdragon 845, and Samsung's Exynos 9820, will match or exceed the Kirin 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
That's pretty absolute. No. "Will match or exceed..." is totally absolute!
And when I quoted you the first time (this is the second) I bolded the problem parts for a reason. Here it is:
"More to the point, Qualcomm's Snapdragon845, and Samsung's Exynos 9820, will match or exceed the Kirin 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
And then you start your explanation by bolding an entire opening paragraph on the Kirin 970!
And in your reply you throw in the Snapdragon 855 (which wasn't even mentioned) and all the absolutes are now nowhere to be seen!
What you are trying to do is re-work what you said. A simple 'my bad' would have been enough. Not even that if you hadn't thrown in the 'better response' and 'there'you go again' parts
At the end of the day you are now admitting you simply don't know. No problem with that. I don't either.
I don't agree with your lead out but as you've repackaged what you are saying to make it clear it's only your opinion, I find it more than valid.
That said, I should point out that the rumours claim the Kirin 980 will include Cortex A77 cores, which, as far as I know, do not exist. It could be a typo but what if it isn't?
How could you form an opinion on cores that no one has any information on?
Don't get me wrong, I'm inclined to think it's a typo but I won't close the door completely on that, just in case.
Okay, so here's one of the reason's that I'm confident about by statement.
First off, there is no A77 core, and the A76 core is 4 wide microarchitecture compared to Apple's current 6 wide architecture.
So this is the prediction that Anadtech has for the comparative performance of the A76 core vs the A11 Monsoon core;
"In terms of general IPC Arm promises a ~25% increase in integer workloads and a ~35% increase in ASIMD/floating point workloads. Together with up to 90% higher memory bandwidth figures compared to the A75 the A76 is then meant to provide around a 28% increase in GeekBench4 and 35% more JavaScript performance (Octane, JetStream). In AI inferencing workloads the doubled ASIMD 128-bit capabilities of the A76 serves to quadruple the general matrix multiply performance in half precision formats.
These performance figures are respectable but not quite earth-shattering considering the tone of the improvements of the µarch. However it’s to note that we’re expecting the A76 to come first be deployed in flagship SoCs on TSMC’s 7nm process which allows for increased clocks."
That pretty much describes the Kirin 980, and the Qualcomm 855, but note, the Kirin 780 is still quite far behind the Qualcomm 845 as is. The A12's microarchitecture and Apple's experience in optimization pretty much guarantees that the A12 will continue holding the crown for both single thread and multiple thread performance.
"I'm confident about by statement"
You may be as 'confident' as you see fit but your 'absolute' comments are now nowhere to be seen except for this:
That pretty much describes the Kirin 980, and the Qualcomm 855, but note, the Kirin 780 is still quite far behind the Qualcomm 845 as is.
"As is"? No. That's impossible. The Kirin 980 doesn't exist!
You are speculating, and wildly at that. "As is" has no place here. Speculation is fine though.
But if you are going to speculate why not at least find a place for those mystery Antutu benchmarks which set a new high (blasting past the 845) and the possibility of a custom GPU which is rumoured to be far faster than the GPU on the 845?
Could it be that that would blow a hole the size of the Grand Canyon in your argument?
I suggest we limit ourselves to speculation or rumours and forget about putting absolutes on the table.
Typo; the 970 is far behind the 845.
If you are arguing that the Kirin 980 doesn't exist, then why were you fine with posting rumors to the benchmarks of that same SOC?
"Of course, AnTuTu doesn't consider CPU performance alone, so this doesn't count for much at the end of the day."
If the latest ARM core is the A76, (there is no A77 yet), and assuming that core was even available early enough for the 980 will be comparable to the 855 architecturally, both at 7nm. As for the GPU, I never stated anything about GPU performance. You might be quite correct that it exceeds both the 845 and the 855 when it arrives, and almost certainly exceeds that A12, but there is still that necessity to stay within the TDP of a smartphone, hence why I don't buy those Antutu benchmarks for the 980 which arrived before the 980 is even in production at TMSC. If you read the Anandtech article, the performance of these A76 cores is based on the theoretical high clock rate of around 3 Gnz, which isn't likely to happen for a production part.
So, I've been talking about the CPU architecture, and you are leaning on Huawei to provide a massive GPU advantage.
There is no conflict in rumours or speculation on something that doesn't exist. That is what speculation and rumours are for.
The difference is when you make a direct factual reference to something that doesn't exist which is what you did but you have said it was a typo. That's fine.
Also, take into account the timing. The Kirin 970 existed about five months before the Snapdragon 845 even shipped and as I said earlier, was based on an older but more mature design.
We are talking about SoCs so it's logical to include GPUs, modems, DSPs etc.
Although it doesn't exist and we can't make absolute claims, as we are clearly close to a release (Huawei has stated a new processor will be revealed at IFA Berlin), a lot of information can be gleaned from the different certification processes.
From that and to give just one example, it looks like the balong 765 might form part of the 980.
"More to the point, Qualcomm's Snapdragon845, and Samsung's Exynos 9820, will match or exceed the KKiri 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
Interesting that you seem so confident.
You might have an answer in around five weeks from now:
"Rumors have it that for the first time Huawei may pack its own graphics processor inside it. The performance of the GPU of the Kirin 980 is speculated to be 1.5 times better than Adreno 630 that comes with Snapdragon 845."
We'll see how that plays out.
Benchmarks are one thing - sustainable performance is another thing which NOT many tech websites dig deep into, except AnandTech. AnandTech had praised the Kirin 950 on the sustained performance and thermal management 2 years back. But they criticized the recent Kirin 960 and 970. We will have to wait for a deep dive by AnandTech to see how Kirin 980 holds up.
Most definitely.
At this point all we have is rumour and speculation. I just wanted to counter what tmay has already presented as 'absolutes' (although he has absolutely nothing to back the claims up) with the only information that is really out (marking it clearly for what it is (rumours).
If they present their own GPU (something that has been heavily suspected for nearly two years now) it will be first generation and possibly lacking optimizations. On the other hand, the 7nm process will help with efficiencies overall and the second generation NPU is rumoured to be much more capable per watt than the first gen product. Added to that, one of the tentpole features of Turbo GPU is its up to 30% power saving but all that is to be seen.
There you go again, adding to the speculation in favor of your favored Huawei.
How about you just sit back and wait for the 980 to arrive in a product, and the "Turbo GT" software to be released and available for review and benchmarking?
Every major SOC maker will be a 7nm this fall, so it will be easy to compare performance amongst the new devices.
No. Speculation and rumour is all their is. Nothing more. That's why I made a point of making that clear when you jumped in feet first with your 'absolute' comments that aren't based on anything that is really known, real or independently scrutinised
You say 'this' or 'that' is this or that way but you simply don't know. No one does!
Yes, sit back and wait. That's the correct way. Mention rumours and speculation if you wish but don't start throwing absolutes around because there are none (in the context of what matters here).
Absolutely, the Kirin 970 does not do as well in benchmarks as the A11 or the Qualcomm 845; it isn't even close.
Qualcomm has it's 855 that will a a 7nm SOC coming out later than the others, but otherwise, I'm not seeing the Kirin 980 as making such a large jump with off the shelf compute cores. You may be right that it will do better than the 845 in graphics performance, but that isn't the end all be all of performance benchmarks. Hence why I stated that the competition will match or exceed the Kirin 980.
That is better from you but wasn't it you yourself who said I needed a better response?
This is what you actually said:
"More to the point, Qualcomm's Snapdragon 845, and Samsung's Exynos 9820, will match or exceed the Kirin 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
That's pretty absolute. No. "Will match or exceed..." is totally absolute!
And when I quoted you the first time (this is the second) I bolded the problem parts for a reason. Here it is:
"More to the point, Qualcomm's Snapdragon845, and Samsung's Exynos 9820, will match or exceed the Kirin 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
And then you start your explanation by bolding an entire opening paragraph on the Kirin 970!
And in your reply you throw in the Snapdragon 855 (which wasn't even mentioned) and all the absolutes are now nowhere to be seen!
What you are trying to do is re-work what you said. A simple 'my bad' would have been enough. Not even that if you hadn't thrown in the 'better response' and 'there'you go again' parts
At the end of the day you are now admitting you simply don't know. No problem with that. I don't either.
I don't agree with your lead out but as you've repackaged what you are saying to make it clear it's only your opinion, I find it more than valid.
That said, I should point out that the rumours claim the Kirin 980 will include Cortex A77 cores, which, as far as I know, do not exist. It could be a typo but what if it isn't?
How could you form an opinion on cores that no one has any information on?
Don't get me wrong, I'm inclined to think it's a typo but I won't close the door completely on that, just in case.
Okay, so here's one of the reason's that I'm confident about by statement.
First off, there is no A77 core, and the A76 core is 4 wide microarchitecture compared to Apple's current 6 wide architecture.
So this is the prediction that Anadtech has for the comparative performance of the A76 core vs the A11 Monsoon core;
"In terms of general IPC Arm promises a ~25% increase in integer workloads and a ~35% increase in ASIMD/floating point workloads. Together with up to 90% higher memory bandwidth figures compared to the A75 the A76 is then meant to provide around a 28% increase in GeekBench4 and 35% more JavaScript performance (Octane, JetStream). In AI inferencing workloads the doubled ASIMD 128-bit capabilities of the A76 serves to quadruple the general matrix multiply performance in half precision formats.
These performance figures are respectable but not quite earth-shattering considering the tone of the improvements of the µarch. However it’s to note that we’re expecting the A76 to come first be deployed in flagship SoCs on TSMC’s 7nm process which allows for increased clocks."
That pretty much describes the Kirin 980, and the Qualcomm 855, but note, the Kirin 780 is still quite far behind the Qualcomm 845 as is. The A12's microarchitecture and Apple's experience in optimization pretty much guarantees that the A12 will continue holding the crown for both single thread and multiple thread performance.
"I'm confident about by statement"
You may be as 'confident' as you see fit but your 'absolute' comments are now nowhere to be seen except for this:
That pretty much describes the Kirin 980, and the Qualcomm 855, but note, the Kirin 780 is still quite far behind the Qualcomm 845 as is.
"As is"? No. That's impossible. The Kirin 980 doesn't exist!
You are speculating, and wildly at that. "As is" has no place here. Speculation is fine though.
But if you are going to speculate why not at least find a place for those mystery Antutu benchmarks which set a new high (blasting past the 845) and the possibility of a custom GPU which is rumoured to be far faster than the GPU on the 845?
Could it be that that would blow a hole the size of the Grand Canyon in your argument?
I suggest we limit ourselves to speculation or rumours and forget about putting absolutes on the table.
Typo; the 970 is far behind the 845.
If you are arguing that the Kirin 980 doesn't exist, then why were you fine with posting rumors to the benchmarks of that same SOC?
"Of course, AnTuTu doesn't consider CPU performance alone, so this doesn't count for much at the end of the day."
If the latest ARM core is the A76, (there is no A77 yet), and assuming that core was even available early enough for the 980 will be comparable to the 855 architecturally, both at 7nm. As for the GPU, I never stated anything about GPU performance. You might be quite correct that it exceeds both the 845 and the 855 when it arrives, and almost certainly exceeds that A12, but there is still that necessity to stay within the TDP of a smartphone, hence why I don't buy those Antutu benchmarks for the 980 which arrived before the 980 is even in production at TMSC. If you read the Anandtech article, the performance of these A76 cores is based on the theoretical high clock rate of around 3 Gnz, which isn't likely to happen for a production part.
So, I've been talking about the CPU architecture, and you are leaning on Huawei to provide a massive GPU advantage.
There is no conflict in rumours or speculation on something that doesn't exist. That is what speculation and rumours are for.
The difference is when you make a direct factual reference to something that doesn't exist which is what you did but you have said it was a typo. That's fine.
Also, take into account the timing. The Kirin 970 existed about five months before the Snapdragon 845 even shipped and as I said earlier, was based on an older but more mature design.
We are talking about SoCs so it's logical to include GPUs, modems, DSPs etc.
Although it doesn't exist and we can't make absolute claims, as we are clearly close to a release (Huawei has stated a new processor will be revealed at IFA Berlin), a lot of information can be gleaned from the different certification processes.
From that and to give just one example, it looks like the balong 765 might form part of the 980.
"More to the point, Qualcomm's Snapdragon845, and Samsung's Exynos 9820, will match or exceed the KKiri 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
Interesting that you seem so confident.
You might have an answer in around five weeks from now:
"Rumors have it that for the first time Huawei may pack its own graphics processor inside it. The performance of the GPU of the Kirin 980 is speculated to be 1.5 times better than Adreno 630 that comes with Snapdragon 845."
We'll see how that plays out.
Benchmarks are one thing - sustainable performance is another thing which NOT many tech websites dig deep into, except AnandTech. AnandTech had praised the Kirin 950 on the sustained performance and thermal management 2 years back. But they criticized the recent Kirin 960 and 970. We will have to wait for a deep dive by AnandTech to see how Kirin 980 holds up.
Most definitely.
At this point all we have is rumour and speculation. I just wanted to counter what tmay has already presented as 'absolutes' (although he has absolutely nothing to back the claims up) with the only information that is really out (marking it clearly for what it is (rumours).
If they present their own GPU (something that has been heavily suspected for nearly two years now) it will be first generation and possibly lacking optimizations. On the other hand, the 7nm process will help with efficiencies overall and the second generation NPU is rumoured to be much more capable per watt than the first gen product. Added to that, one of the tentpole features of Turbo GPU is its up to 30% power saving but all that is to be seen.
There you go again, adding to the speculation in favor of your favored Huawei.
How about you just sit back and wait for the 980 to arrive in a product, and the "Turbo GT" software to be released and available for review and benchmarking?
Every major SOC maker will be a 7nm this fall, so it will be easy to compare performance amongst the new devices.
No. Speculation and rumour is all their is. Nothing more. That's why I made a point of making that clear when you jumped in feet first with your 'absolute' comments that aren't based on anything that is really known, real or independently scrutinised
You say 'this' or 'that' is this or that way but you simply don't know. No one does!
Yes, sit back and wait. That's the correct way. Mention rumours and speculation if you wish but don't start throwing absolutes around because there are none (in the context of what matters here).
Absolutely, the Kirin 970 does not do as well in benchmarks as the A11 or the Qualcomm 845; it isn't even close.
Qualcomm has it's 855 that will a a 7nm SOC coming out later than the others, but otherwise, I'm not seeing the Kirin 980 as making such a large jump with off the shelf compute cores. You may be right that it will do better than the 845 in graphics performance, but that isn't the end all be all of performance benchmarks. Hence why I stated that the competition will match or exceed the Kirin 980.
That is better from you but wasn't it you yourself who said I needed a better response?
This is what you actually said:
"More to the point, Qualcomm's Snapdragon 845, and Samsung's Exynos 9820, will match or exceed the Kirin 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
That's pretty absolute. No. "Will match or exceed..." is totally absolute!
And when I quoted you the first time (this is the second) I bolded the problem parts for a reason. Here it is:
"More to the point, Qualcomm's Snapdragon845, and Samsung's Exynos 9820, will match or exceed the Kirin 970 and 980, and I'm betting that the A12 will as well, so I'm not seeing much of a race."
And then you start your explanation by bolding an entire opening paragraph on the Kirin 970!
And in your reply you throw in the Snapdragon 855 (which wasn't even mentioned) and all the absolutes are now nowhere to be seen!
What you are trying to do is re-work what you said. A simple 'my bad' would have been enough. Not even that if you hadn't thrown in the 'better response' and 'there'you go again' parts
At the end of the day you are now admitting you simply don't know. No problem with that. I don't either.
I don't agree with your lead out but as you've repackaged what you are saying to make it clear it's only your opinion, I find it more than valid.
That said, I should point out that the rumours claim the Kirin 980 will include Cortex A77 cores, which, as far as I know, do not exist. It could be a typo but what if it isn't?
How could you form an opinion on cores that no one has any information on?
Don't get me wrong, I'm inclined to think it's a typo but I won't close the door completely on that, just in case.
Okay, so here's one of the reason's that I'm confident about by statement.
First off, there is no A77 core, and the A76 core is 4 wide microarchitecture compared to Apple's current 6 wide architecture.
So this is the prediction that Anadtech has for the comparative performance of the A76 core vs the A11 Monsoon core;
"In terms of general IPC Arm promises a ~25% increase in integer workloads and a ~35% increase in ASIMD/floating point workloads. Together with up to 90% higher memory bandwidth figures compared to the A75 the A76 is then meant to provide around a 28% increase in GeekBench4 and 35% more JavaScript performance (Octane, JetStream). In AI inferencing workloads the doubled ASIMD 128-bit capabilities of the A76 serves to quadruple the general matrix multiply performance in half precision formats.
These performance figures are respectable but not quite earth-shattering considering the tone of the improvements of the µarch. However it’s to note that we’re expecting the A76 to come first be deployed in flagship SoCs on TSMC’s 7nm process which allows for increased clocks."
That pretty much describes the Kirin 980, and the Qualcomm 855, but note, the Kirin 780 is still quite far behind the Qualcomm 845 as is. The A12's microarchitecture and Apple's experience in optimization pretty much guarantees that the A12 will continue holding the crown for both single thread and multiple thread performance.
"I'm confident about by statement"
You may be as 'confident' as you see fit but your 'absolute' comments are now nowhere to be seen except for this:
That pretty much describes the Kirin 980, and the Qualcomm 855, but note, the Kirin 780 is still quite far behind the Qualcomm 845 as is.
"As is"? No. That's impossible. The Kirin 980 doesn't exist!
You are speculating, and wildly at that. "As is" has no place here. Speculation is fine though.
But if you are going to speculate why not at least find a place for those mystery Antutu benchmarks which set a new high (blasting past the 845) and the possibility of a custom GPU which is rumoured to be far faster than the GPU on the 845?
Could it be that that would blow a hole the size of the Grand Canyon in your argument?
I suggest we limit ourselves to speculation or rumours and forget about putting absolutes on the table.
Typo; the 970 is far behind the 845.
If you are arguing that the Kirin 980 doesn't exist, then why were you fine with posting rumors to the benchmarks of that same SOC?
"Of course, AnTuTu doesn't consider CPU performance alone, so this doesn't count for much at the end of the day."
If the latest ARM core is the A76, (there is no A77 yet), and assuming that core was even available early enough for the 980 will be comparable to the 855 architecturally, both at 7nm. As for the GPU, I never stated anything about GPU performance. You might be quite correct that it exceeds both the 845 and the 855 when it arrives, and almost certainly exceeds that A12, but there is still that necessity to stay within the TDP of a smartphone, hence why I don't buy those Antutu benchmarks for the 980 which arrived before the 980 is even in production at TMSC. If you read the Anandtech article, the performance of these A76 cores is based on the theoretical high clock rate of around 3 Gnz, which isn't likely to happen for a production part.
So, I've been talking about the CPU architecture, and you are leaning on Huawei to provide a massive GPU advantage.
There is no conflict in rumours or speculation on something that doesn't exist. That is what speculation and rumours are for.
The difference is when you make a direct factual reference to something that doesn't exist which is what you did but you have said it was a typo. That's fine.
Also, take into account the timing. The Kirin 970 existed about five months before the Snapdragon 845 even shipped and as I said earlier, was based on an older but more mature design.
We are talking about SoCs so it's logical to include GPUs, modems, DSPs etc.
Although it doesn't exist and we can't make absolute claims, as we are clearly close to a release (Huawei has stated a new processor will be revealed at IFA Berlin), a lot of information can be gleaned from the different certification processes.
From that and to give just one example, it looks like the balong 765 might form part of the 980.
There are your Galaxy S10 at 7nm and triple lens rumors. Looks like Huawei will have even more competition first of the year.
That would be great. And let's not forget Apple. When the bar gets set high, it's up to others to up their game and go even higher. More often than not, consumers benefit from the competition through competitive pricing and a nice product spread.
Comments
You may be as 'confident' as you see fit but your 'absolute' comments are now nowhere to be seen except for this:
"As is"? No. That's impossible. The Kirin 980 doesn't exist!
You are speculating, and wildly at that. "As is" has no place here. Speculation is fine though.
But if you are going to speculate why not at least find a place for those mystery Antutu benchmarks which set a new high (blasting past the 845) and the possibility of a custom GPU which is rumoured to be far faster than the GPU on the 845?
Could it be that that would blow a hole the size of the Grand Canyon in your argument?
I suggest we limit ourselves to speculation or rumours and forget about putting absolutes on the table.
https://news.cgtn.com/news/3d3d514e7845444d79457a6333566d54/share.html
Both articles put these achievements into the push for 200,000,000 units.
If you are arguing that the Kirin 980 doesn't exist, then why were you fine with posting rumors to the benchmarks of that same SOC?
Your link from a previous post above.
https://www.notebookcheck.net/Kirin-980-powered-Huawei-Mate-20-makes-shock-AnTuTu-appearance-outperforms-the-Snapdragon-845.296680.0.html
"Of course, AnTuTu doesn't consider CPU performance alone, so this doesn't count for much at the end of the day."
If the latest ARM core is the A76, (there is no A77 yet), and assuming that core was even available early enough for the 980 will be comparable to the 855 architecturally, both at 7nm. As for the GPU, I never stated anything about GPU performance. You might be quite correct that it exceeds both the 845 and the 855 when it arrives, and almost certainly exceeds that A12, but there is still that necessity to stay within the TDP of a smartphone, hence why I don't buy those Antutu benchmarks for the 980 which arrived before the 980 is even in production at TMSC. If you read the Anandtech article, the performance of these A76 cores is based on the theoretical high clock rate of around 3 Gnz, which isn't likely to happen for a production part.
So, I've been talking about the CPU architecture, and you are leaning on Huawei to provide a massive GPU advantage.
The difference is when you make a direct factual reference to something that doesn't exist which is what you did but you have said it was a typo. That's fine.
Also, take into account the timing. The Kirin 970 existed about five months before the Snapdragon 845 even shipped and as I said earlier, was based on an older but more mature design.
We are talking about SoCs so it's logical to include GPUs, modems, DSPs etc.
Although it doesn't exist and we can't make absolute claims, as we are clearly close to a release (Huawei has stated a new processor will be revealed at IFA Berlin), a lot of information can be gleaned from the different certification processes.
From that and to give just one example, it looks like the balong 765 might form part of the 980.
http://www.hisilicon.com/en/Media-Center/News/Balong 765 news
On a side note. It looks like Adreno Turbo will be revealed tomorrow to go head to head with Turbo GPU.
There are your Galaxy S10 at 7nm and triple lens rumors. Looks like Huawei will have even more competition first of the year.