AMD A8-5500 and Intel i5-5670/90k Render Times Comparison?

[MeesterJackson]

Ok, so hold on, yeah yeah, I know the i5 wins, this wasn't a "VS" sort of thing or a "which is better" thread. But more so I just wanted to know how much faster the i5 would actually be. 

 

My current CPU IS the A8-5500 and I'm going to be upgrading soon. Many of you have told me to go X99 based off my needs/future needs. However...I AM coming from a fairly weak/low end CPU (Technically APU, but my old PC DID have a dedicated GPU). So I'd like to know, has anyone got an actual cart with these two CPU's in it for rendering or at the very least overall performance (though I'd prefer a rendering/workstation chart). 

 

Because if the i5 is A LOT faster, then I...actually might just indeed go for an E3 Xeon or i7 (if I feel like it's worth the extra damn 40+ dollars). I mean, my current CPU takes 2 HOURS to render a 15 minute 720p (I cannot stress that enough, NOT 1080p, but YES 720p) video. Now granted, I AM using "Two Pass" on Sony Vegas which improves bit-rate stability and quality, but also at the cost of essentially DOUBLING the render time as it IS essentially rendering the video twice (those of you who use Vegas and use this feature know this already). 

 

But now that, that's been said, just how much faster would the i5 be? Would it cut the rendering in half and make it an hour long (counting Two Pass)? Or would it only chop off 30 minutes? 

 

Anyone know? 

 

Thanks. 

[JefferyD90]

Well because your not really increasing core count...  My best guess will be about 30-40% time difference.

[MeesterJackson]

Well because your not really increasing core count...  My best guess will be about 30-40% time difference.

Any chance you could give me numbers? I don't really like percentages.  

[JefferyD90]

#1 I don't think your numbers from your original post is good numbers.  I think a setting of yours is WAAYYY off.

 

#2 According to your numbers...  2 hours * 60 minutes in a hour = 120 minutes total...   120 minutes * (35%-100%) =78 minutes...  78min/60min in a hour = 1 hour and 18 minutes.

 

 

---  Edit

 

I know, MATH!

[MeesterJackson]

#1 I don't think your numbers from your original post is good numbers.  I think a setting of yours is WAAYYY off.

 

#2 According to your numbers...  2 hours * 60 minutes in a hour = 120 minutes total...   120 minutes * (35%-100%) =78 minutes...  78min/60min in a hour = 1 hour and 18 minutes.

 

 

---  Edit

 

I know, MATH!

1. Nope. I just use Driftor's High Quality YouTube settings. Looks great to me. Remember the A8-5500 is an older APU and also remember that AMD sometimes uses modules for their cores (mainly shown in the FX line up) to make their CPU's cheaper. Again, Using Two Pass DOUBLES the render time. 

 

2. So it would nearly cut it in half? What about the i7/E3 Xeon's and their Hyper-Threading?

 

So I've noticed. lol

[Opcode]

According to this video it takes 1 hour to render a 40 minute 720p video with a stock i7-3770k

 

 

If money is tight and you don't overclock I would target the Xeon's that have Hyper-Threading as the E3-1231 is nearly the same thing as a i7-4770 with just a 100 MHz slower boost clock.

[JefferyD90]

1. Nope. I just use Driftor's High Quality YouTube settings. Looks great to me. Remember the A8-5500 is an older APU and also remember that AMD sometimes uses modules for their cores (mainly shown in the FX line up) to make their CPU's cheaper. Again, Using Two Pass DOUBLES the render time. 

 

2. So it would nearly cut it in half? What about the i7/E3 Xeon's and their Hyper-Threading?

 

So I've noticed. lol

The A8-5500 uses Piledriver cores... thus they are "modular" but these modules have 2 cores each.  The only real shared thing is the L3 cache, which when working on shared workloads is actually better.  This is one of the better APUs that AMD has made.  Its a lower clocked part, probably cause its a lower wattage part.  But the i5 is still going to be a alright upgrade.  If your looking into the i7's a 4770k will maybe look at a 50% difference, and the new Haswell-E part is legitimately the BEST option and you might see 60% differences. 

[Opcode]

The A8-5500 uses Piledriver cores... thus they are "modular" but these modules have 2 cores each.  The only real shared thing is the L3 cache, which when working on shared workloads is actually better.  This is one of the better APUs that AMD has made.  Its a lower clocked part, probably cause its a lower wattage part.  But the i5 is still going to be a alright upgrade.  If your looking into the i7's a 4770k will maybe look at a 50% difference, and the new Haswell-E part is legitimately the BEST option and you might see 60% differences. 

APU's don't have a shared L3 cache due to the sheer size of L3 (takes up way too much die space). I am curious as if the OP has tried rendering with the iGPU built into the APU (e.g. using "Render using GPU if available"). There's 256 VLIW 4 streams packed in the A8-5500 which should speed things up a bit.

 

[JefferyD90]

APU's don't have a shared L3 cache due to the sheer size of L3 (takes up way too much die space). I am curious as if the OP has tried rendering with the iGPU built into the APU (e.g. using "Render using GPU if available"). There's 256 VLIW 4 streams packed in the A8-5500 which should speed things up a bit.

 

Then in a APU the modules don't have any shared resources.  Therefor, its a quad core through and through. 

 

Edit  ----

 

Exactly what I was thinking, who renders with their CPU anymore?

[MeesterJackson]

According to this video it takes 1 hour to render a 40 minute 720p video with a stock i7-3770k

 

If money is tight and you don't overclock I would target the Xeon's that have Hyper-Threading as the E3-1231 is nearly the same thing as a i7-4770 with just a 100 MHz slower boost clock.

Well my kind sir, if you read the post, you would've seen that my original plan was X99 due to all the comments. LOL

 

However, I still think I'd rock a Xeon at this point. It seems like it'll give me a great performance boost (unless OC'ing an i7 is going to save me another 10 minutes or more, which I don't think it will, I'll stick with the Xeon regardless). Actually, doesn't the 4770 have a 3.4Ghz Base and 3.8 Turbo? The 1231 v3 has the same clock speeds. The 1230 v3 has 3.3 and 3.7. As for the 4770k which has 3.5 and 3.9, the 1241 v3 matches that (though in CPUBenchmark the 1246 v3 matches the 4770k better due to the extra features and power from the iGPU in both).

 

If the Xeon makes the 15 minute video in Two Pass render in under an hour, then yes, I'd still take it (which I believe it would be close to or would).

[MeesterJackson]

APU's don't have a shared L3 cache due to the sheer size of L3 (takes up way too much die space). I am curious as if the OP has tried rendering with the iGPU built into the APU (e.g. using "Render using GPU if available"). There's 256 VLIW 4 streams packed in the A8-5500 which should speed things up a bit.

 

It has a dedicated GPU in the build. I'd have to take out the GPU then which I wouldn't want to anyways. 

[MeesterJackson]

Then in a APU the modules don't have any shared resources.  Therefor, its a quad core through and through. 

 

Edit  ----

 

Exactly what I was thinking, who renders with their CPU anymore?

I'm going to have to I'm going Nvidia. Cuda isn't really supported by Vegas as much as Vegas "says it is" It's really not. It still seems to use the CPU unless there was an update to fix this. Though in time I will also use Premiere Pro...probably still not as much as Vegas though. 

[GeathX]

I'm going to have to I'm going Nvidia. Cuda isn't really supported by Vegas as much as Vegas "says it is" It's really not. It still seems to use the CPU unless there was an update to fix this. Though in time I will also use Premiere Pro...probably still not as much as Vegas though. 

 

Yeah Vegas is much more OpenCL orientated, also where most of the industry is moving towards. Though don't worry about going Nvidia, they are picking up their game CL wise with the 900 series.