If you have 7800, buying 845 wont improve things.
You need a new platform.

 

Flanker has a thread on it with some benches, we are awaiting some game benches from him
http://www.overclock.net/t/1591886/excavator-desktop-yeaah-there-is-it
Quote:

Originally Posted by hojnikb

If you have 7800, buying 845 wont improve things.
You need a new platform.

I believe he was playing with the integrated graphics

 

Quote:

Originally Posted by 7850K

Flanker has a thread on it with some benches, we are awaiting some game benches from him
http://www.overclock.net/t/1591886/excavator-desktop-yeaah-there-is-it
I believe he was playing with the integrated graphics

Even so, replacing 7800 with 845 won't improve things much.
Both run at the same base frequency. And 7800 has more l2 cache, while athlon 845 has arch improvements.

 

Quote:

Originally Posted by hojnikb

Even so, replacing 7800 with 845 won't improve things much.

He's already ordered it, but thanks for stopping by concerned intel citizen

 

Quote:

Originally Posted by 7850K

He's already ordered it, but thanks for stopping by concerned intel citizen

Just point out the obvious. Some people are expecting some magical perf. improvements with exacator. But sad truth is, it just isn't the case.

 

The x4 845 doesn't have integrated graphics, so I hope you have a video card to go along with that CPU.

Otherwise, what you need is a video card, not a CPU. The x4 845 should be about 10% faster or so than the 7800, which isn't much of a game changer (it is about 15% faster in the FPU department).

We don't, yet, know how this will translate to the gaming department. I have a HTPC running a 7870XT video card and an A8-7600 which I am going to use to compare gaming performance against the x4 845 I ordered, so we will know... some time after the 8th, when they get their stock in and then ship it out.

 

Quote:

Originally Posted by looncraz

The x4 845 doesn't have integrated graphics, so I hope you have a video card to go along with that CPU.

Otherwise, what you need is a video card, not a CPU. The x4 845 should be about 10% faster or so than the 7800, which isn't much of a game changer (it is about 15% faster in the FPU department).

We don't, yet, know how this will translate to the gaming department. I have a HTPC running a 7870XT video card and an A8-7600 which I am going to use to compare gaming performance against the x4 845 I ordered, so we will know... some time after the 8th, when they get their stock in and then ship it out.

I was going to pair the 845 with a Geforce GTX 950 since I have a 4k HDMI 2.0 TV now.

Even the highest end APU's from AMD seem to struggle with fairly optimized games at 720p so I decided to give up on the APU thing. I built a 7700k rig for my friend and he is itching to upgrade after a year because of newer games not running well on it at all. I would get him a Radeon 240/250 but it seems like a waste considering you can just get a more superior dedicated card for like 30 bucks more.

Even though the 845 is weaker I am not trying to push anything extreme gaming wise. I just want some IPC improvements and I like a 65w envelope.

70 for 845
plus
130 for 950

200 bucks for a solid mid range CPU-GPU solution. Radeon APU Crossfire solution is more and performs worse.

Still an AMD lover but it is what it is....

 

For connecting to a TV, I'd much prefer the R7 370 if I had to buy new:

http://www.newegg.com/Product/Product.aspx?Item=N82E16814127886&cm_re=r7_370-_-14-127-886-_-Product

However, I'd try to find a used 7870XT (not the non-XT!) or 7950.

A 240 is pretty pointless if you have an AMD APU already, so your friend, and the 250 is not much better. A good 7870 would be a better choice, IMHO.

 

He wants the HDMI 2.0 from the 950 so why would a 370 be better for that?

He probably was looking at the 240/250 for crossfire with the APU which he realizes sucks.

I'm sure he knows a 7870 would be better than a 250..

 

Quote:

Originally Posted by Horsemama1956

He wants the HDMI 2.0 from the 950 so why would a 370 be better for that?

He probably was looking at the 240/250 for crossfire with the APU which he realizes sucks.

I'm sure he knows a 7870 would be better than a 250..

Ah, yes, HDMI 2.0 - only required for 4k @ 60hz, which he isn't going to power with any of the video cards mentioned.

CrossfireX is completely not worth it when you can just buy a faster video card from the beginning.

 

Quote:

Originally Posted by looncraz

Ah, yes, HDMI 2.0 - only required for 4k @ 60hz, which he isn't going to power with any of the video cards mentioned.

CrossfireX is completely not worth it when you can just buy a faster video card from the beginning.

Maybe he wants HDMI 2.0 for 4k video playback.

 

I had a R9 285 but sold it to a good guy on here so he could run proper crossfire. Was a great card.

4K TV was just 300 bucks and Club3d was out of adapters and I normally have good experience with Geforce as long as it is priced properly which it was, 130 vs 170.

I am fine with running 1440p in all honesty but 1080p/720p struggled with the APU but it was still impressive for the value it was, just not strong enough for what I am wanting to do going forth.

I will post benches when the 845 comes in. I can't help it I am a AMD SKU whore. I gotta try everything they release lol

 

Quote:

Originally Posted by IvantheDugtrio

Maybe he wants HDMI 2.0 for 4k video playback.

HDMI 2.0 isn't required for that. Video playback is 30hz, and HDMI 1.2a can do that just fine.

 

Quote:

Originally Posted by dlee7283

I had a R9 285 but sold it to a good guy on here so he could run proper crossfire. Was a great card.

4K TV was just 300 bucks and Club3d was out of adapters and I normally have good experience with Geforce as long as it is priced properly which it was, 130 vs 170.

I am fine with running 1440p in all honesty but 1080p/720p struggled with the APU but it was still impressive for the value it was, just not strong enough for what I am wanting to do going forth.

I will post benches when the 845 comes in. I can't help it I am a AMD SKU whore. I gotta try everything they release lol

Wow, that's a steal! I'm a hardware whore - especially SSDs - I have five in my computer, and every computer in my house has an SSD - save for one I am building for someone else (GTX 960, i5 2500k, etc.). I also love thermal management, I have stacks of fans and heatsinks, and I run custom water-cooling in my main rig (not for graphics, though).

I ordered an x4 845 as well, so I guess you and I represent 25% of the pre-orders, and another user on Anandtech forums, with whom I have been conversing, takes us to 37.5% of pre-orders, LOL! Soon, I will have collected you all!

 

Some user is uploading some benchs from the 845 here:

https://www.reddit.com/r/Amd/comments/48wk6q/just_got_my_amd_athlon_x4_845_here_are_some/
http://imgur.com/a/DTsNz

I made at comparison at the CPU-Z bench, with my 7700K stock (3.4GHz base and 3.8GHz Turbo):


It's 10% more IPC, single and multi (the score is 14% higher because it has 100MHz more baseblock)

 

Would love to see some gaming benchmarks for it, sometimes the optimizations on these chips only show when tested this way. Take things like Planetside 2 and other CPU busting games.

 

Quote:

Originally Posted by Scougar

Would love to see some gaming benchmarks for it, sometimes the optimizations on these chips only show when tested this way. Take things like Planetside 2 and other CPU busting games.

I will be doing exactly that, but my interests will be at 3GHz performance for direct IPC comparisons with several other architectures.

I've already completed the Sandy Bridge 3GHz testing, and am about 1/3 done with K10 (Phenom II) testing, and I've completed the basic layout of the interrogation for simple online consumption. I'll finish with K10 and Steamroller today, then try to get my old Core 2 machine up and running - and at 3GHz - so I can do a Penryn comparison, which is a good way to weed out some of the instruction-set performance effects that skew results somewhat.

 

Got mine in today!

Also for those updating BIOS code within Windows 10 for Carrizo please just flash old fashion way in bios instead. Gigabyte's Windows 7 tool works in Windows 10 but crashes midway through flash.

 

Quote:

Originally Posted by dlee7283




Got mine in today!

Also for those updating BIOS code within Windows 10 for Carrizo please just flash old fashion way in bios instead. Gigabyte's Windows 7 tool works in Windows 10 but crashes midway through flash.

Sweet! Mine should be in tomorrow. Mind running a couple quickie benchmarks at 3GHz so we can compare numbers?

GeekBench (need the integer ST/MT, floating ST/MT, and memory ST/MT subscores)
The CPU-Z benchmark
Cinebench R10, R11.5 (use both in their 32-bit versions)
Cinebench R15 64-bit
SuperPi (1M & 2M)
Dolphin (time in seconds)
7-Zip 32M 1/4thread
3d Particle Movement (single & multi-threaded).

Any you'd like to do would be awesome.

In addition, running Firefox 44.0.2 and running Peacekeeper and JetStream would be quite helpful for comparison.

Again, everything locked at 3Ghz, and DDR3-1600 CL9, ideally.

As many as you'd care to do would be great - I took the time to list them in my order of preference.

 

Dolphin(Lower Better)
Skylake Pentium- 8.51
AMD 7800B- 14.17
Athlon x4 845-12.53

3D Movement(Higher Better)

Single Thread
Skylake Pentium 110.3867
A10 7800B-70.7649
Athlon X4 845 70.8895

Multithread
A10 7800B-183.4777
Skylake Pentium 179.5078
Athlon X4 845-149.0813

Also here is my Geekbench 3 score for the X4 845

http://browser.primatelabs.com/geekbench3/5555662

based on my tests it appears the lack of level 2 hurts the X4 845 in multithread

It is basically on par single thread wise with Kavari with 4MB Level 2 vs Carrizo 2MB Level 2

Dolphin seems to run better on Intel platforms, Carrizo showed improvement in that benchmark over Kavari.

all tests were done at 2133 speed on both platforms, CQ C6 and Turbo were all enabled as test results were worse without it.

There seems to be some small improvements in some areas that are canceled out in others by the lack of level 2

Basically it appears this seals the deal as the Bulldozer arch was complete failure and there were no fixes to make it significantly better. The Pentum G4400 Skylake basically performs better with 2 cores and 3.3 Clock vs 3.7 Turbo on the AMD's.

Skylake, low end board is pretty much the same cost as Carrizo/ low end board but not even apu graphics of any kind on the X4 845.

 

Quote:

Originally Posted by dlee7283

Dolphin(Lower Better)
There seems to be some small improvements in some areas that are canceled out in others by the lack of level 2

There's little doubt the smaller L2 will cause some issues for direct comparisons, but I have some methods to weed out that effect somewhat.

It seems that there may be throttling with the x4 845 in stock form (I know my A8-7600 does it), which is one of the reasons why 3Ghz fixed clockrate results are more useful for comparing generational changes. My interests lie in determining a likely performance profile and level for Zen, as the level should be based off Excavator and the profile should be closer to the same as Phenom II or Sandy Bridge.

From my particular collection of benchmarks, it appears that Steamroller is about 35% slower in integer and 55~60% slower in floating point than Sandy Bridge, but I haven't actually built the worksheet to do the math for me yet, I'm just collecting data

I updated the motherboard BIOS (for Carrizo support) and am, oddly, getting slightly higher results now, though, so I have to see if that holds up through more benchmarks.

 

Quote:

Originally Posted by dlee7283

Got mine in today!

Also for those updating BIOS code within Windows 10 for Carrizo please just flash old fashion way in bios instead. Gigabyte's Windows 7 tool works in Windows 10 but crashes midway through flash.

That heatsink looks familiar. Can you tell if it has a round copper center or not?
I have an early FX-4100 heatsink that has the copper center and in my opinion it works very well for a stock heatsink. That one looks exactly the same, except it has a red fan.
Anyway, I'm just curious. This is an example of what I'm talking about: http://www.tomshardware.com/forum/329180-28-4100-heatsink

 

Quote:

Originally Posted by gaster

That heatsink looks familiar. Can you tell if it has a round copper center or not?
I have an early FX-4100 heatsink that has the copper center and in my opinion it works very well for a stock heatsink. That one looks exactly the same, except it has a red fan.
Anyway, I'm just curious. This is an example of what I'm talking about: http://www.tomshardware.com/forum/329180-28-4100-heatsink

It seems to be almost identical (I have both), but the new one is a bit heavier - I think it may have a bit more copper. The fan is an AVC DSSC0715R2M, with hydraulic bearing. Quite a good fan, really, compared to the old Cooler Master FA07015L12LPB they usually use. The AVC fan is *not* any larger than the old fan, though, and has the same fins (count, pitch, & size). It's just the motor, PWM circuit, body, bearing, and color that are different

And there's no stupid sticker on the low pressure side of the hub (the one you see).

 

Okay, so I have quite a few results now, and I made an interesting observation - about Steamroller's performance and BIOS updates.

I first ran all of my benchmarks on my A8-7600 without modifying my system at all - same old BIOS, same old drivers, everything. I then flashed the BIOS to the BIOS with Carrizo support. I reran all benchmarks - and there was quite a jump in performance at times (as much as 13% for multi-threaded memory performance!). I think this may be due to updated code provided by AMD to BIOS vendors - and may be one of the reason why we are seeing such a wide array of performance numbers for Excavator. I will provide all these numbers, in chart form, in my upcoming interrogation of Excavator.

Now, my old BIOS version for the FM2A88M-Extreme4+ was 2.40, from 8/15/2014. I updated to 2.90 (2/19/16) for Carrizo support. After doing so, via the UEFI, I restarted into Windows 7, the system reloaded every driver as if I had just moved the installation into a new computer - it was a drastic update, it seems. Then, today, when I installed the x4 845, and did absolutely nothing else, it happened again. Very strange. It even reloaded the graphics drivers, PCI drivers, network, USB... literally, everything.

My early results with Excavator suggest a notably more capable memory controller than Steamroller, and a not-so-consistent ~9% or so IPC advantage. I am also seeing much reduced power usage at 3Ghz (I have yet to actually do any full-speed or gaming testing).

The performance profile, though, DEFINITELY suggests that the reduced L2 capacity is hurting performance - particularly in multi-threaded scenarios. Higher frequency should help negate this more than it helped Steamroller, so I suspect Excavator may have better scaling with frequency.

 

On DDR3 side the IMC in Excavator is exactly the same as on Steamroller. Only the combo side differs, GDDR5 on Steamroller and DDR4 on Excavator.