多种电脑平台的音频性能对比
相信对近期准备升级电脑的朋友会有帮助的。
反正我看完是彻底放弃AMD了。
转自 http://www.adkproaudio.com/benchmarks.cfm
[size=+1]
Benchmark tests of Nuendo, Cubase SX, Sonar and others for various computer sub-systems.
This page will be updated from time to time.
These tests were designed to give a good idea of what to expect from various computer subsystems.
They are not designed to compare our systems with the competition. We have no need to boast. (although feel free to download and test your system)
Each test is downloadable and duplicable for independant verification. *NOTE* this is not for marketing our products, but simply to give you a better idea of what each sub-system will deliver and to help you better decide on which platform is best for you and your budget.
Testing Methodology:
Each test was done in 44.1, 24bit unless otherwise specified.
Each test was run at least 2 times with the highest numbers used (worst case). Any oddities were re-tested.
The numbers reported are NOT an average but the absolute highest spike in the CPU meters (worst case)
The audio engine is reset between each buffer change.
If there was ANY pops, clicks at a buffer setting, that buffer setting was NOT included in the graphs. The system may have opened it at that buffer.
No Stereo pair audio cards were used, multichannel cards we used EG: RME Multiface, Fireface, Madi, or M-Audio.
The more channels the card has, the more CPU it will use.
反正我看完是彻底放弃AMD了。
转自 http://www.adkproaudio.com/benchmarks.cfm
[size=+1]
Performance tests for Digital Audio Workstations
Benchmark tests of Nuendo, Cubase SX, Sonar and others for various computer sub-systems.
This page will be updated from time to time.
These tests were designed to give a good idea of what to expect from various computer subsystems.
They are not designed to compare our systems with the competition. We have no need to boast. (although feel free to download and test your system)
Each test is downloadable and duplicable for independant verification. *NOTE* this is not for marketing our products, but simply to give you a better idea of what each sub-system will deliver and to help you better decide on which platform is best for you and your budget.
Testing Methodology:
Each test was done in 44.1, 24bit unless otherwise specified.
Each test was run at least 2 times with the highest numbers used (worst case). Any oddities were re-tested.
The numbers reported are NOT an average but the absolute highest spike in the CPU meters (worst case)
The audio engine is reset between each buffer change.
If there was ANY pops, clicks at a buffer setting, that buffer setting was NOT included in the graphs. The system may have opened it at that buffer.
No Stereo pair audio cards were used, multichannel cards we used EG: RME Multiface, Fireface, Madi, or M-Audio.
The more channels the card has, the more CPU it will use.
| N3 Magneto Test laptops Same test as below for desktops. We included the E6600 numbers for a comparison of how well laptops do again desktops. All tests were preformed with an RME Fireface 800. The ADK laptops are equiped with onboard Texas Instruments Firewire. The object of the test is to load as many Magneto plug-ins without pops clicks in playback. Download here[/td] |
| ADK Sonar 6 Producer Test laptops Same test as below for desktops. We included the E6600 numbers for a comparison of how well laptops do again desktops. All tests were preformed with an RME Fireface 800. The ADK laptops are equiped with onboard Texas Instruments Firewire. If you take these numbers (aside from the 128 buffer) there is a 30% increase in ability of the E6600 over the T7200. The E6600 is 33% faster CPU wise. so had the laptops been equiped with an equal to the E6600 they would have been dead even with a desktop! Download here |
| ADK N3 Magneto Test This test is based on the demo song "Set Me free" found on the Nuendo 3 DVD. We used an RME Fireface on all systems. Firewire adds .5 -.75ms latency The object of the test is to load as many Magneto plug-ins without pops clicks in playback. Download here |
| ADK Sonar 6 Producer Project: 20 audio tracks - each has EQ enabled and VC64 7 instances of Pantheon reverb applied directly on trcks 1-7 8 busses with VC64 on each, Pantheon reverb on reverb bus Session Drummer2 live with all outputs and MIDI playing Pentagon1 and Roland Groove Synth actively playing with MIDI tracks 2 audio loops running Loop Explorer open RME Fireface 800 used on all tests *** the quad cores showed an odd cpu load where 1 of the 4 or 8 cores would have a higher load than the others. like 20% higher. The AMD 6000 played @ 128 buffer with 66% but had alot of pops/clicks. Download here |
| Nuendo 3 (Thonex Test (made by a fellow Nuendo forumite) ) 3 reverb A, 11 Magnetos, 7 Multiband compressors, 1 a1, 3 Embracers, 5 LM-7s, 4 Monologues, 1 Vb-1, 2 DaTubes, 2 Dynamics, 2 Nuendo EQ2s, 1 Q, 2 Step Filters, 1 Tonic, 2 Vovoders, 3 Karlettes. This test also works with Nuendo 2 and Cubase SX 2 and 3. (have to replace 1 EQ in Cubase) Download here Need the Karlette also! The Xeon was tested with a Multiface, Opteron with a Madi, the others a M-Audio Firewire 410. ***See comments below Updated 3-24-06 | |
| Nuendo 3 (Fudd Test) This is a very heavy hitting project and is the closest to "Real World" test we have. This project contains: Blofeld returns Demo, and numerous things we added: Midi- 12 midi tracks (all going to a VSTi), Audio - 38 audio tracks (pre-recorded), 32 mono audio tracks (routed to all inputs) (VST Connections 16 mono inputs/ 8 Stereo Outputs), Vsti- 3 Absynth 3.01 demos, 1 kontakt (4patchs), 2 monologues, 1 embracer, 1 A1, Effects- 3 FX channels, 2 group channels, 74 channel EQ's 4bands enabled on all pre-recorded, Audio, VSTi outputs, FX Channels, and Group Channels, Insert Effects- 1 Tonic, 6 magnetos, 3 VST Dynamics, 2 Reverb A, 3 Roomworks, 2 Chourus, 1 Double Delay, 1 Quadrafuzz, Send Effects - 1 Chorus, 1 Reverb A, 1 Double Delay This test is not that hard to duplicate, requires downloading our file (with instructions) and 2 Native Instruments Demos. Download here | |
| Sonar 4 (Scott Reams (a fellow Sonar Forumite) ) The file contains: 19 audio tracks~ (No actual audio), 3 DXi audio tracks , 3 MIDI tracks (routed to the DXis), 5 Busses (3 setup as auxes) , 19 Track EQs (4 bands enabled for each), 6 Sonitus Gates , 6 Sonitus Compressors, 1 Sonitus Delay , 1 Sonitus Multiband, 1 Sonitus Reverb , 2 Lexicon Pantheon Reverbs, 2 Dreamstation instances (each setup for 16-note polyphony), 1 Cyclone instance There is no actual audio in this test so if the system would open it at a buffer and not drop the audio engine we used the CPU meter reading. Download here | |
| Sonar 4 (Gary with audio) An updated version of Scott Reams Sonar test. WE added numerous items to the test. 22 tracks total plus 5 Buses (shows as 25 tracks but cause of how sonar lists it’s really only 22), 19 of which are audio then 2 Dream Stations and one Cyclone, The plug in breakdown goes like this:, Sonitus Gate: 21, Sonitus EQ: 22, Sonitus Compressor: 23, Sonitus Multiband: 1, Lecicon Pantheon: 24, Midi Stuff came out to… Echo Delay: 3, Transpose: 3, Velocity: 3. then add an MP3 as an audio track. any will do. This way you have audio to determine pops and clicks. Download here (you will need to add your own audio track) Updated 3-24-06 | |
| Comments: the Thonex test was updated on 7-24-06 adding Conroe and woodcrest the Motherboard used for the AMD FX 60 is one of the newer PCIe boards. the older tests were performed on an Nforce 3 AGP. you can see the differance the newer board makes for AMD on the 64buffer. the older X2 4400 wins over the 400Mhz faster FX 60. again more proof that the PCIe boards are not performing as well as the older AGP. It also shows that the Intel processor with 4 Meg Cache and 1066 FSB coupled with DDR2 800 is closing the Gap. **Sonar 4 Gary; the AMD FX60 @ 128 buffer had some very minor clicks. we tested using the RME Fireface. we also tested with the FW-410 and higher CPU load but played clean. based on what we seen lately we blame this on the fireface and not the FX 60. so i felt it needed to be included. The Intel 955 did really well against AMD in this test. |
原帖由 白勺 于 08-2-25 14:51 发表
我觉得性能价格比来说，6600应该是无敌的。毕竟比phenom 9500贵不了多少。但是性能还能更高一些。
而且还可以装黑苹果。。。。。
如果不超频的话，q6600的性能价格比要比9500低10%左右，而且9500跑音频基本也只比Q6600差10%。
超频另当别论~~~~~Q6600 3.2G跟玩一样，9500现在的步进只能稳定到2.8。
不过9500马上也要出黑盒了，另外搭配AMD自己的主板，可以做到四个核心分别调倍频，比如一个核心2.4，另外一个体质好的3.0，剩下两个2.8……
等黑盒降到999的时候，马上抓个790FX配着玩~~~~~~~~~
苹果那就没法说了，不过现在用sonar的人也越来越多~~~~~~~~
[ 本帖最后由 macubex 于 08-2-25 14:57 编辑 ]