Forum Discussion
Honored ContributorCyclone V vs Cyclone IV static power
From Quartus compilation I see that:
EP4CE115F29C7 115k LE typical 98.46 mW max 130.41 mW 5CEBA5U19C7 77k LE typical 227.17 mW max 322.46 mW But... Cyc V shouldn't be 40% lower static power????14 Replies
Altera_ForumBy the way, I agree with nicejob, based on the existing observation, it seems like Cyclone V will have higher power consumption than Cyclone IV. If Cyclone IV can meet your design requirement, probably you can use Cyclone IV to have better power performance.
- Altera_Forum
Sound confusing from the service request. Currently what we are seeing here is significant static power increase from Cyclone IV to Cyclone V. I assume that since under worst case, we already have static power reduction, typical case there should be more power reduction. Just wonder why in typical case we see power increase here. Seems to contradict with the response provided by service request. Probably this is something worth highlighting to them.
- Altera_Forum
I get another response from Service Request (n. 11160246):
Static power is dependent on density and temperature. The power reduction is based on worst case condition using higher temperature. .. this seem a ridiculous excuse.. Quartus can do only typical and maximum PowerPlay analysis.. how can a customer benefit from worst case? - Altera_Forum
hmmm...interesting...
are you migrating your design from CiV to CV? There are some truth in the statement though, I wouldnt bother migrating just for the 30% (even if it is true) considering the price difference and the effort re-assigning my pins to match my existing board. - Altera_Forum
Probably the info in the web was kind of outdated as per the service request response.
- Altera_Forum
Here is what support answered me:
Thank you for using mySupport. The higher power in Cyclone V is expected due to the additional features as compared to CycloneIV. User are recommended to use CycloneIV if their design could fit well into CycloneIV and not needing the additional features. I hope this answer your question. Regards, Very bad answer for -30% static power marketing guys.. - Altera_Forum
I see static power -30% and dynamic power -50%
http://www.alteraforum.com/forum/attachment.php?attachmentid=10845&stc=1 Only to sell devices? But then static is +300% and dynamic maybe -50% ? - Altera_Forum
Hi Flz.
From my experience in running power analysis on Cyclone V vs Cyclone III/IV. The static power is higher in Cyclone V, but the dynamic power was significantly lower. So there was a wash in or improvement in power if the total number of LE was about the same. I wasn't using the low power version of the Cyclone IV in my testing (1.0 V) core, so you need to keep that in mind if you are using the 1.0V core. Static power at smaller geometries always get worse, because leakage currents go up. Although they have improved the material and at 14 nM they are going to a different geometry to combat this issue, it still holds true, it's just not as bad as it could be... However, unless you design is just sitting idle the entire time, look at the dynamic power as well. I think the marketing term of 40% savings, has a disclaimer if you read the fine print that it's related to dynamic power.. Dynamic power is basically follows the formula Power=CV^2F so reducing the core voltage is a big win for power. You can also reduce C in smaller geometries if your traces are short because the transistors are closer together. The F term, Frequency is assumed is the same, but the benefit of smaller geometries, you can switch the transistor faster as well. Pete - Altera_Forum
I sent a service request but I expect nothing...
28 nm LP aren't so good for power consumption... device are hotter! Only I/O thermal power dissipation is better. With Cyclone V heatsink + fan for serious design are a must! Anyone know if story is same on Xilinx side? - Altera_Forum
It seems like the core Cyclone V 110k static power is ~ 4x of the Cyclone IV 115k. This seems rather far from the 40% lower power. You might want to file a service request to Altera to seek further clarification.
