ATI 5730 GPU Temps.

Has anyone figured out how to get temps off the 5730 yet? Why are the current programs out not picking up readings ? :confused2:

 

What about this?
http://foro.noticias3d.com/vbulletin/showpost.php?p=3222545&postcount=148

HD5650 in Acer Aspire 5740G from Spanish forum.
Temps measured while OCed to 650MHz/800MHz making it effectively HD5730. Frankly seems to be too low to be true.

 

If that is the actual case...we should be able 2 rock the 5730 lol. That temp does seem low.

 

Meh, that's a 40nm card with a pretty low TDP of about 20W. My bet is that these readings are accurate.

 

Yeah, it's 26W- HD5730 that is. But I thought you tested AMD GPU Clock and it was unable to read temperature of this GPU?

 

Did you tried the latest version of AMD GPU Clock Tool (v0.9.26.0)?

PS: Everest v5.30.2009 beta also read the temps. So latest beta (5.30.2018) should work.

 

OK, while we're at it. Can someone please explain it to me?
I thought HD5730 was just a higher clocked HD5650 but that would mean maximal TDP of both should be the same as max clock of 5650 are the same as default clocks of 5730. But TDP differs.

The difference is here:
5650:
# Memory clock speed: 800 MHz DDR3/GDDR3
# Memory data rate: 1.6 Gbps DDR3/GDDR3
# Memory bandwidth: 25.6 GB/sec

5730:
# Memory clock speed: 800 MHz DDR3/GDDR3
# Memory data rate: 3.2 Gbps DDR3/GDDR3
# Memory bandwidth: 25.6 GB/sec

So what is different? Same memory type, same freqs, both are 128bit and have the same bandwidth. How come data rate is different and what does it mean in practical terms?

 

Obviously that's a typo. 3.2 Gbps would be true with GDDR5 memory. 1.6 Gbps is the real memory data rate if we know that memory is running at 800 MHz and it's GDDR3 128 bits.

Also, I think that maximum speed for 5650 is a typo too. It says 450-650 MHz (too vague) for the core, but the truth is Ati delivers the GPU at 600/800 MHz (core/memory). You can see it for example here, in the 5650 that comes with Deviltech DTX.

 

I can agree it's a typo with memory but not with frequencies.
Compal usually overclocks their cards a bit- the same was the case for Compal KHLB2 (first notebook with HD4650) and now in case of Compal NBLB2 (aka DevilTech), however both Acer Aspire 5740G and 5942G have core freqs set at 550MHz for their HD5650s.

 

It just makes no sense to me to make 2 redundant GPUs. That's why I think it must be a typo (the 650 core frequency I mean).

I don't know Compal's policy. It must be that 5650 is delivered at 550/800? That would make sense with that 19 watts TDP.

 

Yeah, but that was an Engineering Sample, so that's a different case

 

Frequency may be set on demand (via GPU BIOS) so Compal just chose to get it higher. Acer did the same with 5940G (4650 @600MHz core) and now put HD5650 in the same model calling it 5942G but with lower core freq (550MHz).

My doubts towards 5650 and 5730 were based on TDP. It seems to be the same card and max freqs of 5650 are 650/800 just like those of 5730.
So one would expect max TDP to be the same however max for 5650 is 19W and TDP for 5730 is said to be 26W

 

And in spanish 5940G memory is also downclocked (750 MHz)

But it's just crazy to make two identical twins gpus like 5650 and 5730. It would be easier to just skip 5650 or 5730 and, on demand, downclock it if vendor wants a lower clock. That 450-650 frequency window is too vague IMO. I don't think they have differents TDP at the same frequency.

 

That's my point. If it's the same GPU why does it have different TDP at the same frequency? I mean I understand typos and all but how many of them can you fit in like 10 lines of text for each GPU.

 

There must be a typo in one of those lines. If not in the 450-650 freq., it must be in the 15-19 W. If they made one with the memory data rate they can do it again. Maybe there was an intern writing that day

 

More likely 5730's TDP is wrong- that would be another mistake in the same specification.
Have you seen ATI commercial with monkey upgrading graphics card?
I wouldn't be surprised if the monkey had been promoted to tech writer

 

So y'all are saying the possibility of overclocking the 5730 might not happen ?

 

No. Worst case scenario is TDP for 5730 is just as stated on ATI website.
Best case scenario is- there is a typo in 5730 specs and in fact TDP is the same for HD5650 with 650/800 and HD5730 650/800. So if 5650 specs are right it would be 19W.

 

if the 5730 is at 19W...how much more could it be pushed ? up to 26W ? Depends on temps ?

 

If 5650 really tops at 62*C only problem for overclocking would be core voltage not temperature. Same goes for 5730 which could possibly go higher than 5650 as it may already have higher core voltage.

 

LOL

Hmm, looking again at specs, 5470 says "15 Watts (GDDR5) or 13 Watts (DDR3)". Well, then, looking at 5770 specs, it has a TDP of 30W (650/800 frequencies, same as 5730 but GDDR5). If 5730 has a TDP of 26W with same frequencies but DDR3, can't we say with GDDR5 it should be of ~30W (more or less)? Then the 5730's TDP should be ok, and 5650's wrong.

 

Thanks for all the input Downloads...I am getting my N61JQ-A1 next week. I had IC Diamond Thermal Paste put on CPU / GPU. Cant wait to run it through some test.

 

I've read, by the way, in the same spanish forum as the one you posted before, that 5730 in idle (Asus N71) was reading high 50s-low 60s. Maybe Everest's wrong reading, too high I think.

 

http://www.amd.com/uk/products/notebook/graphics/ati-mobility-hd-5700/Pages/hd-5730-specs.aspx
http://www.amd.com/UK/PRODUCTS/NOTEBOOK/GRAPHICS/ATI-MOBILITY-HD-5700/Pages/hd-5650-specs.aspx
Maybe there's no typo in the Memory Data Rate - since memory bandwidth is OK? Maybe they're using different memory chips?

 

5470 has 64bit memory so it may different in terms of heat generation than 128bit memory

OK but how do you achieve twice the data rate with the same type of memory, bandwidth and frequencies?

 

No, 5730 and 5650 both use DDR3/GDDR3 memory. It's a typo, memory data rate is just Memory clock speed *2 if it's GDDR3 or *4 if it's GDDR5.

 

You're talking about memory bandwidth, not memory data rate.

 

I know, i know, that's why I said a TDP of around 30W. Don't you think 19-30 is too much difference only because of the use of GDDR5?

 

My guess is that the difference is in details like memory timings and latencies. The latter might me for instance 0,5ns for the 5730 and 1ns for 5650. And lower timings and latencies could result in higher power consumption. But that's only my guess.

 

No, I'm talking about memory data rate, just see yourself:

Ati mobility radeon 5850. It says:

# Memory clock speed: 1 GHz GDDR5 or 900 MHz GDDR3/DDR3
# Memory data rate: 4.0 Gbps GDDR5 or 1.8 Gbps GDDR3/DDR3

Just do your maths.

 

Well, if there's a typo in 5730 specs, then there's also one in the 5830 Coincidence?

 

And where's the typo in the 5830 specs?

 

Memory data rate: 1.6 Gbps DDR3/GDDR3
http://www.amd.com/US/PRODUCTS/NOTEBOOK/GRAPHICS/ATI-MOBILITY-HD-5800/Pages/hd-5830-specs.aspx

 

But it's ok. 800 MHz *2 = 1.6 Gbps

 

And we are exactly where we stared when I asked about that
Nothing adds up.

 

In short, we can deduce there's a typo in the 5650 specs. I'd say the typo its in the core frequency range of 450-650. I'd say it's maybe 450-550, and TDP is 15 for 450 and 19 when it's clocked at 550. And that's that

Don't know, it makes more sense than 650/800 and 19W (and a core freq window of 200 MHz is too much i think).

 

That is possible but you can as well assume that there is a typo in 5730's TDP.
And what about Compal with 600MHz core? Did they overclock it on their own?

 

Yeah, sorry 'bout that, I was in a rush - silly me

I meant 5730 - http://www.amd.com/UK/PRODUCTS/NOTEBOOK/GRAPHICS/ATI-MOBILITY-HD-5700/Pages/hd-5730-specs.aspx these 3.2Gbps is clearly another typo - aside from the clock speeds.

EDIT. But these clock speeds doesn't seem to by a simple typo since the processing power of 5650@650MHz exactly matches the 5730 one. Could be a mistake, but definitely not a typo. This really doesn't add up

 

Could be a typo as well in 5730, but its TDP makes more sense if we compare with 5750 & 5770 TDPs. The only suspicious TDP is the one of 5650 gpu.

You said they did it before with Ati 4650, and they did in the same fashion (from 550 to 600, assuming that's what they've done with 5650).

 

OK, so if that's a typo how come the TDP difference? 7W is too much for just timing/latency change but it's exactly the same difference between GDDR3 version of HD5850 and HD5830. Again 7W with almost the same frequencies (100MHz difference in memory)
I guess we lack some engineering skills. Where is Dilbert when you need him?

EDIT:

Yeah but Acer also sold HD4650 with core clock of 600MHz besides we don't know what range of freqs where set for 4650. It may have been just like for HD5650 or similar which would make both Compal and Acer inside ATI set values.

 

Well, that's what we said 4 pages before

 

But you've to consider that they're not giving a range for 5850 GDDR3 TDP. They just say: "31W". Ok, 31W at 500 MHz or at 625 MHz? Instead, they do it in the proper way giving a range for GDDR5 tdp: "30-39 Watts". But the one for GDDR3 is just a...... cheat.

Just my two cents.

 

In fact, it has to be 31W at 625 core freq, just by deduction. How? Let's see the facts:

- We know that 5850 at 500 MHz (GDDR5) has 30W of TDP.
- Then it's not possible that 5850 GDDR3 has a TDP of 31W at 500 MHz, because it can't be higher than the GDDR5 version.
- So, it has to be 31W at 625 MHz.

Ok, now the next question is: has the same TDP 5830 than 5850? Let's see:

- We know that 5830 at 500 MHz has a TDP of 24W.
- We also know that GDDR5 raises ~8W at that frequency range (5850 GDDR5 at 625 MHz is 39W while 5850 GDDR3 at 625 MHz is 31W).
- Then we can deduce that 5850 at 500 MHz should have a TDP of ~24W (31 minus ~7) (7, not 8, because of the lower frequency, less heat, etc. that worths that 1 watt less).

So the logic says that 5800 series specs are ok. 5700 series are almost ok (5730 memory data rate typo). Then 5650 are the faulty ones

 

There is another argument for faulty frequencies for HD5650 you mentioned.
HD5750 with 550MHz core and HD5770 with 650MHz core (both the same memory) differ by 5W in TDP

Difference of 200MHz between HD5650 450MHz and HD5650 650MHz can't be worth only 4W.

EDIT: And this way we made sense of HD5730. This card would be worthless if 5650 could have been 650/800.
There would be no difference between best of 5650s and 5730. Seems we already found two mistakes- core clocks for 5650 probably should be 450-550. Wonder if anyone else noticed

 

Yes, that monkey promoted to tech writer ....

We should write them. At least they should correct the memory data rate of 5730. Maybe they correct "something" in the 5650 specs too...

 

I am glad my thread helped you guys figure something out...that no one else has??? LOL.....

 

Ressurecting this thread.

I've messed up with my new N71JQ and the Ati 5730. So far, im currently doing gpu tests with encoding Videos with mediashow espresso.

I still have to test in games but so far, i've set my core to 750 and memory to 900.
The gpu is 100% stressed from the encoding and temps are stable at 71C. Ill do more test with actual video benchmarks but this looks good

Any other N71 N61 owners have experimented with OC'S ?

Jov

 

Hello,

i oc'd my hd 5730 from stock settings (650/800) to 750/950 stable. I can go all the way up to 780/1000 stable, but i don't use such high frequencies...

So using 750/950 after some intensive gaming i get temperatures (using also a notebook cooler) of:

i5 520M CPU; max 75°C
HD 5730; GPU max 74°C

You can monitor GPU temperatures with the newest stable RealTemp program, also with HWMonitor - but use the drivers from the ATI site:

Win 7 32-bit:
ATI Catalyst? Mobility Display Driver

Win 7 64-bit:
ATI Catalyst? Mobility Display Driver

Regards,

K

 

How has the gaming performance been?

 

Please note that voltage differences can cause LARGE TDP changes between similar clocked chips.